Genetics, insurance and participation: How a Citizens’ Jury reached its verdict

Genetics, insurance and participation: How a Citizens’ Jury reached its verdict

ARTICLE IN PRESS Social Science & Medicine 64 (2007) 2487–2498 www.elsevier.com/locate/socscimed Genetics, insurance and participation: How a Citize...

174KB Sizes 0 Downloads 1 Views

ARTICLE IN PRESS

Social Science & Medicine 64 (2007) 2487–2498 www.elsevier.com/locate/socscimed

Genetics, insurance and participation: How a Citizens’ Jury reached its verdict Paul Bennetta,, Susan J. Smithb a

University of Edinburgh, Edinburgh, UK b Durham University, UK Available online 5 April 2007

Abstract Access to life insurance is subject to health discrimination. Whether this discrimination should take into account the results of predictive genetic tests is a topic of public debate. This paper draws from the deliberations of the Edinburgh (Scotland) Citizens’ Jury on Genetic Test Results and Life Insurance to evaluate the capacity of one participatory research method to inform debates on genetics and insurance. We show that through a process of knowledge-building, crossexamination and deliberation, ‘lay’ jurors are able to assimilate complex information, engage in subtle argument and arrive at well-reasoned, clearly warranted conclusions. The Citizens’ Jury approach has a further key advantage: it embraces the formation and articulation of normative ideas. It brings public understandings of how things ought to be into an arena dominated by ‘expert’ opinion. These normative indicators, which in this case relate to fairness and trust, are often overlooked by policy makers. However, they are key to democratic decision-taking and relevant for health promotion. r 2007 Elsevier Ltd. All rights reserved. Keywords: Citizens’ Jury; Life insurance; Health discrimination; Genetic testing; Public policy

Introduction There has been a steady displacement of social safety nets by private insurances as governments look to markets to manage a wide range of social, financial and environmental risks (Armstrong, 2005; Ericson, Barry, & Doyle, 2000; Burchardt and Hills, 1998). Although life insurance has always been provided privately (with the exception, in the UK, of a small ‘death benefit’), it too has become increasingly important as a means of protecting households as welfare is restructured. In fact, Corresponding author. Tel.: +44 131 650 2518.

E-mail addresses: [email protected] (P. Bennett), [email protected] (S.J. Smith). 0277-9536/$ - see front matter r 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.socscimed.2007.02.029

British households now spend more on life insurance (excluding pensions) than on any other form of private insurance, paying premiums worth over £29 billion in 2003 (ABI, 2004). For individuals, access to life insurance is explicitly about health discrimination, at least in an actuarial sense. This contrasts with social insurance, which works on the principle of solidarity, wherein everyone takes responsibility for everyone else’s risks as well as their own. In the solidarity model, participation is universal, and contributions are determined by ability to pay rather than risk of claiming. Private insurance, on the other hand, is based on the principle of mutuality. Here participation is voluntary, participants are grouped according to risk, and premiums vary between these

ARTICLE IN PRESS 2488

P. Bennett, S.J. Smith / Social Science & Medicine 64 (2007) 2487–2498

groups reflecting the probability of claims being made (Wilkie, 1997). Calculating this probability demands some kind of risk discrimination. In life insurance, risk is indexed by life expectancy, and assessed with reference to individuals’ health histories and prospects. Different tactics are used by different insurers, but all engage in a process of premium differentiation anchored on health, as they compete for ‘good risks’ (people unlikely to claim) and try to avoid the adverse selection of ‘poor risks’ (Akerlof, 1970; Katzman, 1988). The question of where to draw the (health) line in insurance underwriting, and how to increase the transparency of this process, is considered more fully elsewhere (Smith & Bennett, 2006). Here we focus on one aspect of health risk assessment and discrimination in life insurance: the use of genetic test results. We take this example, and locate it in the UK, for several reasons. First, despite all the positive implications of developments in bioinformation, these new technologies have the potential to increase the social and financial divide between people whose health is known to be at risk and people with no history or prognosis of illness. Genetic discrimination may effectively create a class of ‘healthy ill’ (Hubbard, 1993), if people who are in good health now are treated differently because they might develop adverse health conditions later. Although health discrimination generally has received surprisingly little attention (a point elaborated in Smith & Easterlow, 2005), genetic discrimination more particularly is widely debated across a range of jurisdictions. Just a taste of the literature raises questions like: is genetic discrimination measurable (Otlowski, Taylor, & Barlow-Stewart, 2003), is it widespread (Nowlan, 2002), how might it work (Miller, Ahern, Ogilvie, Giacomini, & Schwartz, 2005), and what are its consequences (Ellison & Rees Jones, 2002)? In sum, as Parthasarathy (2004, p. 332) observes, ‘whether and how genetic information should be used for insurance or employment purposes is one of the most contentious areas of debate in the genomic age’. In jurisdictions like the UK, where there is a well-developed socialised (national) health service, the most pressing questions concerning genetic discrimination in insurance turn on the market for life insurance (see also Taylor et al., 2004). Second, while important in itself, the question of genetic testing is also a lens through which to consider wider issues of health discrimination in insurance markets. Life insurers have a long history

of using health-risk information, but the genetic testing debate provides a chance to reopen the ‘black box’ of underwriting. In the UK, where there is a moratorium on insurers’ use of genetic test results, most people know little about their personal genetic predispositions. They know whether they smoke, keep fit, or have a history of illness, but most have not been tested for the asymptomatic genetic characteristics that may predispose them to premature death. So the issue of genetic testing, in the context of the UK’s legislative setting, and at a time when debates may be more influenced by principle than by personal circumstance, briefly offers the opportunity to explore how health discrimination in insurance does, could and should work. Third, while there is growing interest in public understandings of genetic research, there is comparatively little social involvement in debates on the use of genetic technologies (though see Bates, Lynch, Bevan, & Condit, 2005), and even less participation in the formation of public policy. However, the ongoing UK debate around the possible use of genetic rest results for life insurance underwriting is uniquely open-ended. The existence of a voluntary moratorium (established in 2001, renewed during 2005, and running to 2011) means that no decision has yet been made; but equally it implies that there will at some point be a positive move either to ban or to endorse the use of genetic test results in insurance underwriting. This provides a space of potential in which to tease out the contribution of a lay public to a complex sciencepolicy issue that is normally left to specialists and experts. Exploring this potential, by reporting on a new initiative in participatory research, is the focus of this paper. The aims of the paper are, then, twofold. First, to consider the strengths and limitations of Citizens’ Juries as a research method for eliciting lay ideas about complex scientific, social and ethical decisions. Second, to examine the deliberations and outcomes of a specific Citizens’ Jury on genetic testing and life insurance. We thus aim both to inform methodological debate in social science and medicine, and to intervene in ethical and policy debates concerning health discrimination in life insurance provision. Genetics and insurance: a participatory approach Genetic testing is producing new information about individuals’ health risks. On the one hand,

ARTICLE IN PRESS P. Bennett, S.J. Smith / Social Science & Medicine 64 (2007) 2487–2498

since insurers already base their premiums on all kinds of information about actual and potential health (family health histories, personal health conditions, health-relevant lifestyles), it seems logical to include genetic risks too. After all, fair and viable underwriting depends on risks being disclosed, however they come to light. On the other hand, since genetic testing comes with so many actuarial uncertainties and presents new challenges across a range of social issues, there is an argument that genetic test results are intensely personal and should, by law, remain private. In short, an imperative of commercial viability is set against an ethic of confidentiality. In the UK there is considerable emphasis on ‘stakeholder’ participation in advancing this debate. For instance, the insurance industry negotiated a moratorium with the government on the use of genetic test results in underwriting, except for cover exceeding £500,000 and even then only if the test is approved by the Department of Health’s Genetics and Insurance Committee (GAIC), itself a committee of diverse stakeholders. Applicants can never be required to take tests for insurance purposes, and at the time of writing only one kind of test result, for Huntington’s disease, is approved for use by insurers. Another group of stakeholders, the Human Genetics Commission (HGC), was set up to advise government policy on the full range of ethical issues arising from genetic science. It involves a widerange of non-governmental ‘expert’ participants, including geneticists, ethicists, doctors, health-interest group representatives and lawyers. It also uses open-meetings and encourages public consultation to widen a policy process that might previously have been the exclusive preserve of government and industry. Nevertheless, the incorporation of lay perspectives and direct ‘lay’ public participation in genetics and insurance debates is limited. It is well established that lay perspectives, and especially normative ideas in the conduct of everyday life, rarely find their way into political decisions, a point developed in Roberts, Smith, and Bryce (1995) and Smith (2005). Moreover, the methods by which such views may be elicited are also often found wanting. Opinion polls may tap into lay perspectives, but while they may be statistically representative they are, in the end, ‘superficial and non-interactive, designed to elicit the uninformed views of the public’ (Coote & Lenaghan, 1997, p. 6). Public

2489

meetings and written consultations, on the other hand, tend to downplay the competence and interest of the general public, often operating with a ‘deficitmodel’ which researchers interested in ethical debates about science and health have found wanting (Wynne, 1991; Roberts et al., 1995; Kerr, Cunningham-Burley, & Amos, 1998). This paper looks to forms of deliberative democracy to avoid superficial measures of public opinion, guard against the bias of private vested interests, and extend due consideration to the wide ranging personal, social and organisational implications of complex decisions. ‘Deliberative democracy, broadly defined, is y any one of a family of views according to which the public deliberation of free and equal citizens is the core of legitimate political decisionmaking and self-government’ (Bohman, 1998, p. 401). Whereas liberal representative democracy prioritises individual preferences and assumes such preferences are defined prior to the political act of voting, deliberative democracy proceeds from the assumption that it is only through deliberation that such preferences are formed (Kupar, 1996, p. 2). At least two common principles of democratic deliberation can be identified. The first is a commitment to justifying views through public reasoning (Cohen, 1996) rather than simply deliberating with oneself. For Dryzek (2003, p. 1) ‘deliberation as a social process is distinguished from other kinds of communication in that deliberators are amenable to changing their judgements, preferences, and views during the course of their interactions’. Fearon (1998) explains why this kind of deliberation is an important preface to decision-taking, noting especially that it overcomes the problem of bounded rationality by pooling individuals’ limited knowledge to enhance the quality of debate. Deliberation also ‘exposes people to others’ point of view and creates a community within which it is much more difficult to ignore their interests’ (Ward, 1999, p. 78); it demands the public justification of opinions (Aldred, 2002) and ultimately ‘relies upon a person’s capacity to be swayed by rational arguments and to lay aside particular interests and opinions in deference to overall fairness and the common interests of the collectivity’ (Miller, 2003, pp. 183–184). Following Habermas (1990), it is through deliberation that people can come to a mutual understanding of their generalisable interests. A second key dimension of deliberative democracy is that of participation, rather than deliberation within an elite body. All citizens should be entitled to participate and participation itself should build a

ARTICLE IN PRESS 2490

P. Bennett, S.J. Smith / Social Science & Medicine 64 (2007) 2487–2498

more active ethos of citizenship (Smith & Wales, 2000). In order for outcomes to be regarded as legitimate, such deliberation must be equal—everyone should have a chance to be involved in deliberation and participate equally, without existing distributions of power shaping their chances to contribute (Cohen, 1989). Participation thus refers not only to initial access to deliberative institutions, but also to equality of participation during deliberation (to prevent vested interests from interfering with the formation of generalisable interests). Once again, Habermas (1990) provides the inspiration: deliberation should take place in ‘ideal speech situations’ spurred on only by the force of better arguments (recognising, of course, that this is a regulative ideal to be aimed for, rather than a perfectly achievable state of affairs). For Smith and Wales (2000) the question remains as to how such ideals of deliberative democracy can be institutionalised. How, in the real world, can everyone participate in a way that is also equal and deliberative? To address this, a number of institutional designs have been proposed. Fishkin (1991), one of the pioneers of deliberative democracy, favours large scale ‘deliberative opinion polls’ or even a national ‘deliberation day’ (Ackerman & Fishkin, 2003) as a means for implementation in which several hundred people (or the entire electorate) spend time deliberating issues in small groups, and questioning experts, before voting. Such strategies allow a large number of citizens to participate and deliberate, although even over one or two days, the depth of deliberation across a range of topics may be limited. Fishkin (1991) also recognises the existence of smaller scale forms of deliberative democracy such as Citizens’ Juries, and Smith and Wales (2000) convincingly argue that Citizens’ Juries should be of interest to deliberative democrats. Only 12–16 people participate in a single Citizens’ Jury, but what this method sacrifices in scale, it makes up for in terms of commitment to equal participation, depth of deliberation and meaningful options for agenda setting. Rather, than assuming a priori welldefined views, Citizens’ Juries allow the public to develop both knowledge and preferences as the Jury unfolds (Kenyon, Nevin, & Hanley, 2003) and so participate in ‘informed political deliberations’ (Smith & Wales, 1999, p. 305). To date Citizens’ Juries have been used only sporadically in the United Kingdom (usually by local authorities for health or planning decisions),

and they are rarely employed as a research technique in the social sciences (although see Kenyon, Hanley, & Nevin (2001) and Ward, Norval, Landman, & Pretty (2003) in the fields of environmental sustainability and valuation). We wanted to explore whether the Citizens’ Jury approach, with its commitment to facilitating lay participation in political debates, could also function as an effective participatory research method for social scientists. While more commonly-used focus groups do allow participants to express warranted concerns (Bates et al., 2005), we hoped a longer Citizens’ Jury would allow a greater depth of deliberation and perhaps even form the basis for normative recommendations. The study The Edinburgh Citizens’ Jury on Genetic Test Results and Life Insurance was convened over three full days in July 2004. This timescale was chosen to enhance the scope for information building, active learning, critical scrutiny and agenda forming deliberation (Coote & Lenaghan, 1997; Stewart, Kendall, & Coote, 1994) among people with no prior experience of the issue. In addition to the research team, who acted as independent moderators, three groups of participants were involved: ‘expert witnesses’ who provided information (a former Chief Underwriter, a specialist in Public Health Genetics and a Researcher of genetic testing and insurance policy); ‘policy advocates’ who defended particular policy directions (from genetics and health interest groups and from the insurance industry); and the Jury itself, which, following scrutiny and deliberation, was charged to reach a ‘verdict’. Jurors were recruited from responses to an initially random sample of 1200 names from the Edinburgh electoral register. A three-day exercise dictates a degree of self-exclusion, and full-time employees were under-represented in the 113 positive responses we received. This pool of potential jurors was subsequently stratified by employment status, age, sex, and housing tenure, to secure a sample of 14 jurors approximately evenly split across categories. This is not to argue that people can only represent those who share their characteristics; rather it is an attempt to secure the diverse resources for deliberation that a wide range of backgrounds delivers (Smith & Wales, 2000). The final Jury comprised six men and eight women; six home owners and eight renters; seven people in paid

ARTICLE IN PRESS P. Bennett, S.J. Smith / Social Science & Medicine 64 (2007) 2487–2498

employment and seven selected from students, homemakers, retired and unemployed people. All but one attended for the full three days; each received a single payment of £150 at the end of the programme to cover their costs. Each day performed a distinct purpose: 1. Knowledge building: ‘Expert witnesses’ presented essential background information, explaining the basics of life insurance underwriting, the science of genetic testing, and the current policy framework, including the moratorium. 2. Cross-examination: A ‘court’ session in which three policy advocates presented their preferred ‘post-moratorium policy model’, before defending this model against questions from the Jury. 3. Deliberation: First, by a panel of witnesses and advocates to an agenda set by jurors; second by the Jury itself leading to a ‘verdict’ in favour of one of three policy models. Jurors were asked to consider three policy options which might follow the expiry of the moratorium on the use of genetic tests by insurers. The policy models, negotiated in advance between the researchers and policy advocates, were: 1. legislate to restrict insurers’ access to genetic test results, 2. extend the moratorium, 3. allow insurers to access and use genetic test results (with a State safety net for those who cannot secure private insurance at affordable rates). The Jury were asked to work towards a verdict for two reasons. First, the verdict itself was intended as an intervention in public debate. Second, the process of negotiating a verdict, as much as the outcome itself, allowed us to examine the development of people’s normative aspirations, and evaluate the potential of the Citizens’ Jury as a participatory research method. The study followed appropriate ethical guidelines, and met the requirements of Edinburgh University’s ethics procedure. The moderation team, of course, designed the initial format of the Jury and selected witnesses, but during the course of the study no attempt was made to influence the issues considered important by the Jury, beyond ensuring that all three models were evaluated. The entire proceedings were audio-recorded, transcribed and analysed by a

2491

method of close and repeated reading. These data were secured and are published with participants’ explicit consent. The findings are tabulated to show the full range of ideas contained in these qualitative materials. Quotations are used in the text to emphasise the most insistent themes and to pick out the threads of key arguments. How the Jury reached its verdict The arguments The first day of the jury revolved around a series of short, factual presentations by the ‘expert witnesses’. The aim here was to provide the jury with the information and resources they would need to examine their case. On day two, the main arguments for and against each model were set out by three expert advocates. A close reading of the advocates’ transcripts identifies the eight points of distinction which were drawn between the models, and these are summarised in Table 1. First, each transcript contains a core theme. Model one positions genetic testing for insurance as the thin end of a growing wedge of unfair genetic discrimination which can only be halted by legislation. Model two makes a positive case for ‘sitting on the fence’, extending the moratorium on the grounds of its flexibility, until the issues are more clearly understood. Model three argues for the importance of maintaining the same high standards of actuarial discrimination across all kinds of health risk information, genetic or otherwise. Second, the models can be distinguished by their approach to the evidence base underpinning the reliability and predictive power of genetic test results. The argument for extending the moratorium (model 2) is that the science is uncertain and the evidence incomplete: We don’t know which direction the future’s going to goy And the one thing about prediction in scenarios in an uncertain world is that you can be absolutely sure that you’ve got it wrong. The counter argument for legislating sooner rather than later (model 1) is that the best evidence is already known, and the predictive power of future tests will be less not greater than those currently to hand (so the risk of adverse selection is low): The genes that have been found already are the more predictive ones, so that we know that the

ARTICLE IN PRESS 2492

P. Bennett, S.J. Smith / Social Science & Medicine 64 (2007) 2487–2498

Table 1 Summary of the three advocates’ arguments Argument

Model Legislate to restrict insurers access to genetic test results

Extend moratorium

Allow insurers access to genetic test results, with adequate welfare provision

Main argument in favour of model

Insurance is the thin end of the genetics wedge; at-risk groups need protection. Legislation is a win-win option, with benefits for insurers, the insured and society at large

The moratorium works for now and provides space for rational debate; the future is uncertain. The evidence base is insufficient and may change radically; there are positive reasons for ‘sitting on the fence’

Insurers should be regulated to apply the same high standards to all aspects of health risk assessment. There will always be exclusions, so innovations social insurance are a priority

Evidence base

Geneticists are sceptical of the claims made for predictive tests; the public are vulnerable to ‘quackery’. Future research will not make genetic risk assessment more reliable, just more complex

Innovations in scientific evidence will shape the future in uncertain ways. More evidence also needed on the number and results of genetic tests for cost calculations

The predictive power of genetics is less than other (health) risk indicators. This is unlikely to change, though decisions may be influenced by a range of other clinical and scientific innovations

Genetic exceptionalism

Genetic testing is unique among health tests because genetic make-up cannot be changed. There are precedents for social exceptionalism for political or ethical reasons (ethnic ‘type’ is not used in UK underwriting). Genetic testing could be excepted in the same way.

Genetics and genetic tests may not be exceptional in fact, but the novelty, notoriety and uncertainty that surrounds them warrants special and separate attention for now. It is not feasible or reasonable to treat other health risks in this ‘special’ way

Genetic testing is not the only kind of predictive medical test; there are no grounds for treating it differently.

Viability of insurance industry

There is little evidence of adverse selection; cost concerns are exaggerated

The industry remains viable under the conditions of the moratorium; there is insufficient evidence to assess the implications of the other models

There is an unquestionable financial bottom line, which will lead to exclusions, and which requires social intervention

Prospects for industry self-regulation

There are precedents for insurers misusing health information; there is no reason to assume they will do any better with genetic information

The current process of voluntary regulation works. There whole industry has signed up to a Code. The penalties of noncompliance are prohibitive

There is a problem of trust. A regulatory regime geared to securing the highest standards of evidence for robust decisions on all kinds of risk is required

What people want/ need

Decisions on whether to have genetic tests should not be influenced by concerns about financial exclusion. People who participate in genetic research should benefit from it, not risk laying the groundwork for genetic discrimination

The public demands and deserves insurance at reasonable rates. They have this under the moratorium without having to disclose genetic test results

A form of national insurance is needed for those whose risks are too high for markets to insure at affordable rates

International comparators

A wide range of other countries have legislation preventing insurers accessing genetic test results

Other countries (with legislation) have higher costs and a limited product range

Comparison with Scandinavian welfare systems to illustrate the pros and cons of state provision

Main argument against other models

The moratorium has already worked: not accessing test results protects individuals and has not damaged industry. Access to test results in one areaa (insurance)

Neither of the other models allows for the uncertainties that currently obtain around genetic testing and insurance. Neither keeps the options open.

The moratorium has de-fused debate, but has ‘let everybody of the hook’ by deflecting attention from wider issues around the use of predictive information in risk

It is the fact that conditions are rare, not that they are genetic, that insurers cannot handle

ARTICLE IN PRESS P. Bennett, S.J. Smith / Social Science & Medicine 64 (2007) 2487–2498

2493

Table 1 (continued ) Argument

Model could lead to genetic discrimination in other areas (eg employment)

future tests will be less predictive than the ones we have today. The case for model three—for a level playing field for all health indicators—is that since the predictive power of genetic test results is less than that for other health indicators, there is no need to give them special treatment. Third, a substantial part of the discussion turned on arguments for and against genetic exceptionalism. Here, the case for model one rests on the assumption that genetic conditions are exceptional: I would argue that genetic tests are different from other types of health tests, and that’s mainly because you can’t change your genesy Moreover, this advocate argued that certain other health risk indicators, such as phenotype (a proxy for ‘race’), are already set aside by insurers because they are socially unacceptable. There are, then, precedents for allowing social considerations to outweigh economic and scientific ones without any serious cost to insurers. The argument against this kind of genetic exceptionalism is the basis of model three which asserts that genetic information should not be treated differently from other sorts of health information, provided there are high enough standards and strict enough accountability to ensure actuarially ‘fair’ health discrimination across the board.

Legislation now would be complex, possibly flawed, and too inflexible

assessment. Legislation freezes problematic positions, would be unwieldy and inflexible, particularly if premised on genetic exceptionalism

minority in the form of ‘a mechanism which builds on the solidarity we ought to feel as citizens’. It is precisely the difficulty of choosing between the case for genetic exceptionalism (which uses legislation to prevent insurers accessing genetic test results) and the case for actuarial logic (which demands that all health risk information is treated in the same way) that underpins the appeal of model two. This makes a positive argument for occupying a middle ground by extending the moratorium in order to air all the possibilities and resolve scientific uncertainties. Before we had the moratorium we had lots of companies doing lots of different things. We had lots of those interest groups [discussed earlier] all coming up with different positions and they had a big row about it. We don’t have that any more. We have the kind of debate that we are having here at the moment. A rational debate. The five additional themes set out by the advocates in their position statements, and during the first round of cross-examination, all elaborate these points and are listed in Table 1. The cross-examination

To my view, all predictive information in the context of private, voluntary insurance, should be subject to the same rigorous level of examination. Anybody wanting to use it should be expected to understand [it], and demonstrate competence in understanding it and using it appropriately and fairly.

After hearing the cases for and against three policy options, the jurors began an active process of cross-examination. Initially they worked in small groups to discuss immediate views and generate questions; then the groups pooled ideas and posed questions to the advocates. In this first stage of cross-examination jurors considered each model on its own merits. Some questions were simply points of clarification; the majority were about the winners and losers under each scenario:

This advocate nevertheless insisted that such a system could only work if governments extended some degree of financial security for an excluded

Is it not lopsided for the legislation [ in model 1] to benefit the patient at the expense of the insurance industry?

ARTICLE IN PRESS 2494

P. Bennett, S.J. Smith / Social Science & Medicine 64 (2007) 2487–2498

The status quo works [in model 2] for the insurance company but does it work for the small minority we’ve heard about who are caught up in this? Who is going to pay for, and who would receive help from, a state safety net [in model 3]? These questions signal a shift in the jury’s mode of operation from learning to critical evaluation. To build on this, the Jury were again split into small groups and asked to summarise and prioritise the arguments they found most compelling in favour of each model (Table 2). It is striking that the main advantages attached to each model are features likely to protect the public against financial exclusion, whether by outlawing genetic discrimination, keeping an open mind, or enhancing state safety nets. All the advantages connect with some notion of fairness. Jurors were also asked to summarise the main concerns raised by each model, and to warrant these concerns (also in Table 2). Most concerns centre on the risks each model carries of promoting social inequality, exclusion and insecurity. Finally, in this cross-examination, the Jury worked in groups to produce questions for a mixed

panel of advocates and expert witnesses. Here, opinion-exploring questions not only aired concerns about fairness, but also probed understandings of trustworthiness in relation to both the policy process (where jurors probed the impact of an imbalance of power between stakeholders) and potential policy outcomes (where they raised questions about the extent to which insurers and/or the state could be trusted to act in particular ways once a policy had been agreed): Can insurers be trusted to use genetic test results objectively and in the applicant’s favour; Can the state be trusted to provide the necessary welfare to support those refused private insurance? (model 3) Can insurers be trusted to make the best use of an extension to the moratorium in terms of collecting data to inform future policy? (model 2) Can the law be trusted to be flexible enough to accommodate developments in genetic science? (model 1) Concerns like these indicate that by the time it came to cross-examination, jurors were not simply learning, or making basic comparisons, but were

Table 2 Advantages and disadvantages of the models as identified by the Jury Advantages Model 1 (legislate)  Protects vulnerable groups against discrimination  Protects individuals ‘emotionally, financially and otherwise’  Guarantees the public access to information on policy and practice  Offers an incentive to individuals to present for genetic tests Model 2 (extend moratorium)  ‘Buys time’ to ensure that choices informed by evidence can be made in the future  Protects ‘the majority’ of individuals  Preserves a workable status quo

Disadvantages

 Lacks flexibility to accommodate scientific innovation  May encourage adverse selection and damage the insurance industry

 Valorises inaction (why hasn’t more use been made of the existing moratorium?)

 May allow insurers to ‘hedge their bets’ in their own interests

 Rests on voluntary agreements, increasing the insecurity of people at risk of gene-related health conditions

Model 3 (level health playing field)  Has an openness that works both ways, so that insurers can be challenged  Demands better state safety nets  Provides the basis for fair health discrimination and more cost-effective insurance

 Most health risk information does not affect premiums, so it may be unethical to access it

 There are no guarantees that negative genetic test results



will override other health risk information (e.g. from family history), so the benefits may not outweigh the costs to individuals May compound the social divide between those protected by the market and those reliant on the state

ARTICLE IN PRESS P. Bennett, S.J. Smith / Social Science & Medicine 64 (2007) 2487–2498

critically engaging with the issues. The ‘figure’ of the juror had changed from that of ‘learning subject’ to one of ‘knowledgeable deliberator’. Rather than simply asking questions, jurors began to debate with each other and with the panel. Rather than being content with factual clarification, jurors became increasingly interested in those qualities of the three models which demanded normative judgements. Rather than leaving their values to one side, jurors began to move towards discussing the normative criteria upon which the ‘best’ model could be decided, and it was here that ideas of fairness and trust came to the fore. The deliberations A key finding from this research is that not only can a Citizens’ Jury understand, articulate and warrant plausible concerns about complex social problems, its members can also express and develop normative ideas collectively. That is, we found Citizens’ Juries to be a practical method for developing an understanding of its members’ generalisable interests. So the Jury offered a balanced account of how things are, but it also negotiated, from complex starting points and amid a variety of competing interests, a vision of how things ought to be. The central normative steer from the Jury’s final deliberations hinges on a complex set of ideas about fairness. Model one was depicted as fairest to those most at risk of discrimination: Very definitely, what we’re all agreeing on—we don’t want to see people being discriminated against. I am concerned about legislation because of the lack of informationy but counter to that is also my feeling that legislation is the way forward to stop discrimination. Model two was depicted as most fairly (or least unfairly) balancing the interests of the all stakeholders: It’s always easier to stay with the status quo— absolutely—it doesn’t cause any pain, it doesn’t have cost implications, it doesn’t mean we’ll have to change what we do or how we do it—and that’s good enough. Model three, on the other hand, was recognised as actuarially fair, in treating all health risks in the same way, avoiding the perceived ‘double-standard’

2495

of model 1. It was also deemed fair in an egalitarian sense in that those whose health risks priced them out of the market would have a state safety net. For this reason, jurors began their final deliberations with a clear inclination to model three because of its even-handed treatment of all kinds of risk: Why should [people] be given some sort of special treatment purely because the test is based on genetics and nothing else? I just don’t see the reason why genetic tests should be any different from any other [predictive] tests. They’re not an exact science, and neither are any of the other tests, so why are they different? Fairness was, in the end, the normative yardstick by which the Jury judged all three models. However, as deliberations moved towards a verdict, debate hinged not only on which concept of fairness was best but also on the extent to which the stakeholders could be trusted to enforce it. The verdict, then, came down to question of pragmatism as much as idealism; to an imperative of compliance as much as a goal of social justice. For example, a major stumbling block for Model 3 (which is actuarially fair) was the Jury’s lack of trust in government to provide the social safety nets on which the viability of the ‘level health playing field’ depends. If you were assured that the public sector was going to be stronger, then I could go along with an open system [model 3]. I couldn’t believe [under model 3] the state would come up and help people who are being discriminated against. There is, nevertheless, a clear sense in the Jury’s deliberations that with a stronger welfare ideal in politics, they might opt for Model three: My first choice would have been three with a really good welfare system. I don’t think I’m happy in leaving three; I’m realistic. Turning to Model two, the Jury expressed scepticism at the idea of extending the moratorium early in its deliberations. Although attracted by an element of flexibility and by its potential to balance competing interests, jurors concluded that little of positive value had been achieved in its five years of operation. Indeed, they were so surprised and perplexed by this that the majority of discussion

ARTICLE IN PRESS 2496

P. Bennett, S.J. Smith / Social Science & Medicine 64 (2007) 2487–2498

hinged on how to overcome this inertia should the moratorium be extended: There’s got to be a goal. Some proper steel guidelinesy about the ethics of it, the purposes of it, what they want to actually try and achieve while they are having their moratoriumy’ Ultimately, concerns that the benefits of the existing moratorium had not been reaped made the Jury suspicious of insurers’ motives for continuing it, not least because model two does not ensure compliance (the moratorium is voluntary). Reflecting on the themes of ‘trust’, ‘voluntary’, and ‘massive corporations’ one juror sums up: these three words for me don’t fit together, because I don’t trust the corporations. Model one, with its roots in genetic exceptionalism and its faith in legislation, was passionately and persuasively presented, but it always had a mixed reception. However, a discussion which started with a concern that ‘there is an awful lot of legislation already in place’, ended by agreeing that ‘I’m yet to find a very strong reason really why there shouldn’t be legislation’. The shift of emphasis hinged around the conviction that fairness is ultimately about the extent to which ‘every individual needs to be protected’. To an extent, each model promised this in different ways, but the legislative underpinnings of model 1 were seen to guarantee that the promise was trustworthy: Basically I don’t trust anybody. I trust nobody. Least of all do I trust business; and my trust in the government is fast leavingy Legislation, although sometimes put into place too hastilyy is a ‘belts and braces’ approach to a problem which society really needs to deal with. Model 1 was seen as the only model that bonded actors into a specific form of behaviour, through the concrete sanctions afforded by legislation. It is being bonded into particular forms of action, rather than relying on past experience of trustworthy behaviour, that most underpins trustworthiness (Hardin, 1996). The verdict Thirteen of the fourteen jurors were present for the vote; they did not achieve a unanimous verdict. On a first vote the Jury split 9:4 between Models

One and Two; following a further round of debate, they settled on a split of 10:2:1 spanning all three Models. Throughout the deliberations, there was a normative impulse driving debate and a recognition among jurors that being agreed as to ‘ends’—in this case, an early agreement that people should not be disadvantaged because their health is impaired or at risk—neither dictates the appropriate means, nor guarantees agreement over these. In the end, the Jury’s deliberations turned on a compromise between the ideals of fairness and the pragmatics of trust. All three models were recognised to be ‘fair’ in some respects, in preventing or at least ameliorating the effects of discrimination. The Jury was mindful of the bottom line for insurers, and of the social role of the insurance industry in risk mitigation as well as its economic role as an employer. However, jurors’ sense that they could not completely trust either voluntary agreements with industry or political commitments to social welfare was decisive. Model one was, for these reasons, regarded throughout as the only way to protect individuals against genetic discrimination. It was attractive to the majority of jurors not because it seemed fairest (all the models offered to protect the most vulnerable in different ways) but because it met the criterion of trustworthiness more plausibly than the alternatives. It bonded actors into certain behaviour through the sanction of the law, rather than through voluntary, altruistic codes. The Jury voted for a concept of fairness that protects some, but not all, of those who are most vulnerable, combined with a model of compliance that is enforceable in law rather than rooted in unenforceable conventions of trust. This was chosen as the best practical option rather than as the fairest or most logical of the models. Reflections The participatory research conducted with the Edinburgh Citizens’ Jury provides good grounds for challenging the scepticism of policymakers towards the experiences and expertise of ‘the public’. This is true even, perhaps especially, in complex areas dependent on public understandings of science, or in perplexing (even tedious) subjects around financial services. You do not have to be a member of the Human Genetics Commission or the Genetics and

ARTICLE IN PRESS P. Bennett, S.J. Smith / Social Science & Medicine 64 (2007) 2487–2498

Insurance Committee, nor do you have to be a politician, insurer or NGO, in order to be able to conduct well-informed, practically relevant discussions about a complex ethical and policy issue like genetic testing and insurance. The members of the Edinburgh Citizens’ Jury demonstrated an ability to assimilate information, critically evaluate evidence, and deliberate in a sophisticated fashion, on a complex topic that at first glance may have been thought unlikely to hold their attention. The Citizens’ Jury approach does, moreover, have some advantages over the use of traditional participatory research methods such as focus groups. First, while it allows the incorporation of lay knowledge, it also recognises the value of knowledge-building—to ensure that jurors are equipped with a wide range of perspectives and arguments which they are free to evaluate. Second, the Citizens’ Jury requires the cross-examination of witnesses and advocates; it not only harnesses what jurors know and are told, but also benefits from their ability to determine what they need to know. It allows them to set their own agendas and respects their ability to critically to evaluate competing views. Indeed, whilst the Jury approach begins with the establishment of ‘expert’ witnesses and a ‘lay’ Jury, this expert/lay division is, to an extent, transcended through the course of the proceedings, such that by the final deliberations, the Jury are also experts. Third, the lengthy process encourages Jurors to develop their own normative views. It does not assume that jurors arrive at an issue with clearly defined preferences, on a subject to which they may have given little, if any, previous consideration. Rather it begins with the premise that these values are negotiated through deliberation itself—the very essence of deliberative democracy. There are of course some points of reservation. On a practical note, the exercise is expensive. Neither the expert witnesses nor the policy advocates were paid for their participation. However, travel expenses were paid, as were accommodation and meals, generating a cost of £1700. Together with the £2100 of payments to the jurors, and the costs of mailouts, printing, and transcription, the total cost of the jury came to over £5000, not including the cost of principal investigators, research assistance and estate. More substantively, there are three qualifications of note. First, while the topic of the Jury was not disclosed until the day it convened, some jurors had

2497

more knowledge and experience of genetics and insurance than others. On the one hand, this stimulated debated and allowed sharing of information. On the other, it meant that some jurors—like the wider population—may have already held strong opinions on the debate in hand. We could not be sure that jurors were equally open to reconsidering their opinions on the basis of reasoned argumentation. Second, our method, though participatory in style, did not allow for jurors’ involvement in framing the study as a whole, or in identifying the particular models under scrutiny. Yet some jurors proposed innovative ideas about future policy and this points to the merits of extending the participatory element of Citizens, Juries to harness this originality more explicitly. Third, Citizens’ Juries fall foul of the vexed questions of generalisability and applicability; fourteen people drawn from the electoral register are unlikely to determine the UK’s genetics and insurance future. Furthermore, there is no obvious route for a Jury convened by independent researchers to feed into policy. Although we always made this clear (as part of our ethics procedure) it still raises an ethical question because the act of forming a Jury must imply there is an agenda to influence. A reason for publishing the findings of course is that they show that lay publics do have sufficient competency to engage in complex discussions and reach clearly justified verdicts. This, at least, is a point that to be taken seriously by social scientists and policymakers alike. Acknowledgement This research was funded by the Economic and Social Research Council (Grant no. R000239489). Our thanks go especially to Andrew Tibbs for helping to design and facilitate the Citizens’ Jury. We wish to acknowledge the contribution of four specialist advisors and three policy advocates as well as fourteen jurors to this work. References ABI. (2004). UK long-term insurance net premium income. ABI Statistics, October 2004. Association of British Insurers /http://www.abi.org.ukS. Ackerman, B., & Fishkin, J. (2003). Deliberation day. In J. Fishkin, & P. Laslett (Eds.), Debating deliberative democracy (pp. 7–30). Oxford: Blackwell.

ARTICLE IN PRESS 2498

P. Bennett, S.J. Smith / Social Science & Medicine 64 (2007) 2487–2498

Akerlof, G. (1970). The market for ‘lemons’: Quality uncertainty and the market mechanism. Quarterly Journal of Economics, 84, 488–500. Aldred, J. (2002). It’s good to talk: Deliberative institutions for environmental policy. Philosophy and Geography, 5(2), 133–152. Armstrong, C. (2005). Equality, risk and responsibility: Dworkin on the insurance market. Economy and Society, 34, 451–473. Bates, B., Lynch, J., Bevan, J., & Condit, C. (2005). Warranted concerns, warranted outlooks: A focus group study of public understandings of genetic research. Social Science & Medicine, 60, 331–344. Bohman, J. (1998). Survey article: The coming of age of deliberative democracy. Journal of Political Philosophy, 6(4), 400–425. Burchardt, T., & Hills, J. (1998). From public to private: The case of mortgage payment protection insurance in Great Britain. Housing Studies, 13, 311–323. Cohen, J. (1989). Deliberation and democratic legitimacy. In A. Hamilton, & P. Pettit (Eds.), The good polity: Normative analysis of the state (pp. 17–34). Oxford: Blackwell. Cohen, J. (1996). Procedure and substance in deliberative democracy. In S. Benhabib (Ed.), Democracy and difference (pp. 95–119). Princeton, NJ: Princeton University Press. Coote, A., & Lenaghan, J. (1997). Citizens’ Juries: Theory into practice. London: Institute for Public Policy Research. Dryzek, J. (2003). Deliberative democracy and beyond. Oxford: Oxford University Press. Ellison, G., & Rees Jones, I. (2002). Social identities and the ‘new genetics’: Scientific and social consequences. Critical Public Health, 12, 265–282. Ericson, R., Barry, D., & Doyle, A. (2000). The moral hazards of neo-liberalism: Lessons from the private insurance industry. Economy and Society, 29, 532–558. Fearon, J. (1998). Deliberation as discussion. In J. Elster (Ed.), Deliberative democracy (pp. 44–68). Cambridge: Cambridge University Press. Fishkin, J. (1991). Democracy and deliberation: New directions for democratic reform. New Haven: Yale University Press. Habermas, J. (1990). Moral consciousness and communicative action. Cambridge, MA: MIT Press. Hardin, R. (1996). Trustworthiness. Ethics, 107, 26–42. Hubbard, R. (1993). Predictive genetics and the construction of the healthy ill. Suffolk University Law Review, 27, 1209–1224. Katzman, M. (1988). Societal risk management through the insurance market. In R. C. Hula (Ed.), Market-based public policy (pp. 21–42). London: Macmillan. Kenyon, W., Hanley, N., & Nevin, C. (2001). Citizens’ Juries: An aid to environmental evaluation? Environment and Planning C: Government and Policy, 19, 557–566. Kenyon, W., Nevin, C., & Hanley, N. (2003). Enhancing environmental decision-making using Citizens’ Juries. Local Environment, 8(2), 221–232.

Kerr, A., Cunningham-Burley, S., & Amos, A. (1998). The new genetics and health: Mobilizing lay expertise. Public Understanding of Science, 7, 41–60. Kupar, R. (1996). Citizens’ Juries: The Hertfordshire experience. Working paper (p. 26). University of Hertfordshire Business School. Miller, D. (2003). Deliberative democracy and social choice. In J. S. Fishkin, & P. Laslett (Eds.), Debating deliberative democracy (pp. 182–199). Oxford: Blackwell. Miller, F., Ahern, C., Ogilvie, J., Giacomini, M., & Schwartz, L. (2005). Ruling in and ruling out: Implications of molecular genetic diagnoses for disease classification. Social Science & Medicine, 61, 2536–2545. Nowlan, W. (2002). A rational view of insurance and genetic discrimination. Science, 287, 195–196. Otlowski, M., Taylor, S., & Barlow-Stewart, K. (2003). Genetic discrimination: too few data. European Journal of Human Genetics, 11, 1–2. Parthasarathy, S. (2004). Regulating risk: Defining genetic privacy in the United States and Britain. Science, Technology and Human Values, 29, 332–352. Roberts, H., Smith, S., & Bryce, C. (1995). Children at risk?. Open University Press. Smith, S. J. (2005). States, markets and an ethic of care. Political Geography, 24, 1–20. Smith, S. J., & Bennett, P. (2006). Where to draw the line? Health discrimination in life insurance underwriting. Unpublished manuscript. Smith, S. J., & Easterlow, D. (2005). The strange geography of health inequalities. Transactions of the Institute of British Geographers, 30(2), 173–190. Smith, G., & Wales, C. (1999). The theory and practice of Citizens’ Juries. Policy and Politics, 27(5), 295–308. Smith, G., & Wales, C. (2000). Citizens’ Juries and deliberative democracy. Political Studies, 48, 51–65. Stewart, J., Kendall, E., & Coote, A. (1994). Citizens’ Juries. London: Institute for Public Policy Research. Taylor, S., Otlowski, M., Barlow-Stewart, K., Treloar, S. A., Stranger, M., & Chenoweth, K. (2004). Investigating genetic discrimination in Australia: Opportunities and challenges in the early stages. New Genetics and Society, 23(2), 225–239. Ward, H. (1999). Citizens’ Juries and valuing the environment: A proposal. Environmental Politics, 8(2), 75–96. Ward, H., Norval, A., Landman, T., & Pretty, J. (2003). Open Citizens’ Juries and the politics of sustainability. Political Studies, 51, 282–299. Wilkie, A. (1997). Mutuality and solidarity: Assessing risks and sharing losses. British Actuarial Journal, 3, 985–996. Wynne, B. (1991). Knowledges in context. Science, Technology and Human Values, 16, 111–121.