Tải bản đầy đủ - 0 (trang)
4 Some `Personal' Data Have Public Implications

4 Some `Personal' Data Have Public Implications

Tải bản đầy đủ - 0trang


P. Mills

to the treatment of the person from whom they are collected but also in that they

contribute to scientific understanding, research and development, and provide an

evidence base for policy or administrative decision making. Similarly, failure to use

data in the public interest affects private interests because, as members of the public,

we are severally (in the case of a medical advance and, in particular, in the case of

‘personalised’ medicine) or collectively (in the case of a public health measure)


The Nuffield report draws attention to two questions about public interest: how

to determine the content of the public interest with regard to the use of data and the

force that should be accorded to it, particularly when there are competing individual

interests at stake. In the case of data initiatives in health care addressing this takes

the form of a critical reflection on the conditions of the tension between individual

interests and the ‘other’ of the public interest, namely an examination of the way in

which individual and public interests are mutually implied.

Consequently, our problem is not finding a ‘balance’ between privacy and public interest

for a data initiative, but resolving a double articulation, between the private interest in

protecting privacy and promoting the public good, and the public interest in protecting

privacy and promoting the public good. We all have interests on both sides, private and

public, as individuals, members of families, groups, communities and nations. Navigating

among these different relationships with other individuals, professionals and institutions

requires a subtle negotiation of many different norms of information access and disclosure,

of when and how they may be modified and where hard and fast limits should be drawn.

(Nuffield Council on Bioethics 2015)

This reflection on the mutual implication of privacy and public interest, on

privacy as concerning norms and morally reasonable individual preferences (that

must be morally reasonable when assessed in relation to those norms), and public

interest as the assertion of morally reasonable norms and collective preferences,

leads to the formulation of a question with regard to the moral basis of data

initiatives in general, one that recasts the double bind – which seemed intractable so

long as it embodied a contradiction between imperatives transfixing an actor in the

data protection paradigm – as a co-operative problem, namely: how may we define

a set of morally reasonable expectations about how data will be used in a data

initiative, giving proper attention to the morally relevant interests at stake?5


Finding a use of data that is mutually acceptable to those whose morally relevant interests are

at stake gives rise to the requirement that the expectations about the use of data form a coherent

‘set’, one that does not contain contradictory elements (although surpassed contradictions may be

preserved at levels below that of the collective agreement). Furthermore, respect for those whose

interests are at stake (which, in some cases are protected by rights and entitlements) means that it

must be one that is capable of being articulated (‘publicly statable’) in a way that is meaningful,

and understandable to those whose interests are at stake, such that an account of decisions can be

given to them that they would recognise as reasonable (Daniels and Sabin 1997; Habermas 1990).

Ethical Reuse of Data from Health Care: Data, Persons and Interests


4 Persons and Interests

The approach that the Nuffield report took to addressing this question was to

treat data initiatives as cooperative social practices or, rather, to take seriously the

implications of data sharing being a social practice.6 This means that the report

does not start by asking what kinds of data are in play but rather by asking on what

norms of privacy and disclosure particular uses of data trespass, and what interests

they engage.

In response to the practical question of how to define a set of morally reasonable

expectations the report recognises two main senses in which expectations may be

‘morally reasonable’. The first is as conforming to some independent standard of

moral reasonableness; the second is as the outcome of a process of moral reasoning.7

Rather than offering a purely formal or procedural response, the Nuffield report

made a set of recommendations that embodied elements of both of these approaches,

relying on procedural elements (to ensure a situated outcome capable of accounting

fairly for the interests actually at stake) within parameters set by standards of respect

for persons and human rights (to avoid the possibility of morally perverse outcomes

resulting from a purely procedural approach).

A crucial element of the procedural aspect is participation in the establishment of

the set of expectations by ‘those with morally relevant interests’ in order to put into

play both the expectations and the values associated with them. These agents should

include a full range of interests: not only those who will handle or use the data, or

the outputs of the initiative – the professionals involved – but also those who may

be affected by the initiative, who may stand to benefit or be harmed by different

outcomes, in particular, where the data may be related to them as individuals or

groups.8 The range of such interests and the form in which they precipitate into

distinctive moral agents (individuals, groups, collectives) will differ from initiative

to initiative, which is why there can be no complete universal set of norms, no ‘onesize-fits-all’, as the Nuffield report says. This means that any set of minimal or

general norms, for example those encoded in high level legislation, is unlikely, on

its own, to provide an adequate basis for governance: they may be a poor ‘fit’ for


The Nuffield report remarks, as others have done, on the imprecise and loaded use of language

in the relevant literature. ‘Data sharing’, with its connotations of beneficence and mutuality, is an

example of this although the term seems appropriate in this context where the norms of exclusivity

are established through the involvement of relevant interests to which it restores an appropriate

level of moral agency.


Of course, second order questions arise in each case (for example, in the first case, about the origin

or justification of the ‘independent’ standard; in the second, regarding the fairness of the process).

The report does not try to resolve these but it keeps them in play. Consequently the process is

reflexive and interminable: its conclusions are provisional and persuasive rather than decisive, and

the standard is treated as conventional rather than categorical.


This is not only those from whom the data may have been collected but also those to whom people

the findings might be applied. If the findings inform decisions relating to public policy, this entails

a case for the engagement of the ‘public’ in general (or their representatives).


P. Mills

particular data initiatives, both in the sense of being unnecessarily restrictive and in

the sense of being insufficiently sensitive to the interests involved.

It is through participation that the relevant norms and interests – and the nature of

their relevance – are discovered. In the case of National Health Service patient data,

for example, the scope of morally relevant interests will include the interests of those

treated by and working in the NHS, as well as other potential beneficiaries such as

health policy makers, academic and professional researchers. Through this process,

without denying the existence of conventional rights or duties, what is brought into

question are the circumstances or conditions in which the entitlements would be

claimed or set aside, along with the relevant social norms. But this is not all: it

also, importantly, provides a motor for the transformation of social norms and value

commitments as they encounter each other in a process of respectful deliberation

(Parker 2007).9 This means that moral reasoning need not stop at the threshold

of acceptability – of determining merely what is acceptable – but should have a

continuing and constructive function, moving beyond what is merely tolerable to

what is morally preferable for those whose interests are interdependent. This is

especially appropriate where collective choices have opportunity costs, where the

choice of a suboptimal option may foreclose a better one (Nuffield Council on

Bioethics 2012).

4.1 Establishing Norms, Freedoms and Duties

Three key elements within the set of expectations are the norms of privacy and

disclosure that are at stake in any given data initiative, the form in which respect

for moral agents is expressed, and the requirements of governance in place. These

relate, first, to the common social norms, second, to the freedom of individual moral

agents, and, third, to the duties of professionals, linking the spheres of, respectively,

social relationships, private conduct and professional practice.

To elaborate this scheme I should like briefly to recall the genealogy of this tripartite scheme in the deliberations of the Nuffield Council working party that steered

the development of the report. The scheme grew out of an attempt to categorise

the possible ‘moral bases’ for data sharing and the conditions in which each would

apply. The idea was that these categorical norms would provide a stabilising and

justifying function for claims made about the adequacy and proportionality of the

other measures. The scheme of moral bases originally comprised duty, solidarity,

autonomy, reciprocity, and authority. These were not meant to describe the implied

motivations of all the actors involved in a given data initiative (which are, by


In this respect it has advantages over aggregative approaches that simply gather preferences as

a basis for decision making (e.g. voting) or are simply designed as a take-it-or-leave-it ‘offer’ to

optimise the number or type of those consenting to participate. Nevertheless, owing to the fact that

it is not practical to involve all those who have an interest, it is still necessarily only indicative

(rather than representative).

Ethical Reuse of Data from Health Care: Data, Persons and Interests


hypothesis, confused and inconsistent) but rather the character of relations that

would legitimately prevail at the ‘public’ level. Thus there would be cases in which

it was morally reasonable to require the sharing of data, for example where public

health was at stake, as in the case of an epidemic.10 Hence, while respectful of

individuals, the expression of individual agency in the form ‘opting out’ of data

sharing in such a case would be morally impermissible.

The hypothesis was that the character of this ‘moral basis’ for data sharing would

correlate with the way in which respect is shown for moral agents involved in

the data initiative because of the way in which individual and public interests are

mutually implied. As we proceed through the classification, different moral bases,

by hypothesis, will underwrite different ways of expressing respect for individual

preferences. One significant way in which this is done is through the rather vexed

gamut of practices described as consent, although often less in relation to the giving

of choice and more in relation to the choice given. Thus, in the hypothetical scheme,

the different moral bases would stand as justification for different expectations about

consent, ranging from none, where compliance is a universal duty (or a duty for all

members of a definable class of persons), through implicit consent where there is

a subsisting social or democratic licence, through general and more specific forms

where there are particular requirements because important freedoms are recognised

as being at stake. For example, where the moral basis for data sharing is solidarity

(where there may be a prior acceptance that individuals should be willing to carry

costs on behalf of others) explicit or detailed consent of individuals may not be

required, although individuals may be free to opt out and should not be coerced to

participate (Prainsack and Buyx 2013).11 More specific forms of consent, on the

other hand, become important where the transaction has a more economic character

(where it is a matter of equitable trading off of benefits and costs).

The third element, governance, concerns the assurance of moral conduct of

others: individuals are entitled to have expectations of others using data, particularly

professionals involved in data initiatives. These include expectations about who

these others will be, and how their conduct will be governed. There is a public

interest in ensuring that those involved in data initiatives discharge a moral duty

of care owed to others, especially those whose interests place them in positions

of vulnerability, a duty that is not exhausted simply by complying with subjects’

consent. As with consent, different forms of governance will be appropriate to

different relationships between the public and private interests involved.

In the original approach, norms of privacy and disclosure were foundational. It

was part of the reflection on the various successes and failures of previous data


Interestingly, this case has been made, although not entirely persuasively, in relation to the

sequestration of health data for Big Data analysis supporting biomedical research, health service

planning and delivery (e.g. predictive interventions) and policy. The case is broadly that, without

the gains in efficiency promised by better data use, the National Health Service in England will

become unaffordable and cease to exist, imperilling public health.


Prainsack and Buyx offer a descriptive ethics of solidarity in Solidarity: reflections on an

emerging concept in bioethics (Prainsack and Buyx 2011).

Tài liệu bạn tìm kiếm đã sẵn sàng tải về

4 Some `Personal' Data Have Public Implications

Tải bản đầy đủ ngay(0 tr)