James Wilson, UCL
Care.data sounds good on paper. Commissioning of healthcare services will be more efficient and more accurate. Massive new datasets will drive both research and health improvement. Rare side-effects of drugs will be detected much more easily. Correlations between lifestyle and health will be much easier to spot. The economy will be boosted by increased commercial usage of the datasets. So why did the policy reach such a pitch of crisis that a six month “pause” needed to be declared in February, and what needs to be done now to get it back on track?
Care.data involves the monthly extraction of rich data at the individual patient level from GPs’ computer systems into a central database at the recently established Health and Social Care Information Centre (HSCIC). The data extracted includes directly identifying information such as a person’s age, postcode and NHS number. These identifiers are then replaced with codes in a process known as pseudonymisation. Once the data is in the HSCIC’s secure system, it can be linked to existing databases, such as the Hospital Episodes Statistics database. Identifying data will not exit the HSCIC except in very special circumstances such as a public health emergency.
The Health and Social Care Act 2012 creates an obligation for GPs to supply patient data if NHS England directs HSCIC to collect it. Patients do not have the legal right to opt out of such data collections, but the HSCIC has committed to allow patients the ability to do so. The main reasons for insisting that patients should have the ability to opt out of data collections were ethical: allowing individuals the ability to control information about them follows from a commitment to respect their privacy and autonomy, and helps to maintain individuals’ trust in the NHS.
If the ability to opt out is to provide ethical genuine legitimation for a policy, individuals need to be adequately informed. NHS England did not initially plan a national information campaign for care.data (ruling this out in August 2013), but agreed to one in October. Both the content and the delivery of the resulting leaflet, “Better Information, Better Care” were widely criticised. The leaflet did not mention care.data at all, and no standard form for opting out was produced. Indeed, it became apparent only after the leaflet was circulated that there would be two separate opt out codes, one to prevent your information from leaving your GP’s practice, and the other to prevent information about you from leaving the HSCIC. The leaflet was delivered through the Royal Mail’s door to door service, which is usually used to deliver junk mail, meaning that those who had opted out of receiving junk mail did not receive it, and that most others disregarded it. A poll for the BBC found that only 29% of adults recalled receiving it.
By the time that the six month pause was announced in February, only 19% of GPs supported an opt out approach, with three quarters supporting an explicit opt in to the scheme. However, an eleventh hour conversion to an opt in policy is politically highly unlikely. Shifting to an opt-in system would require considerable further delays in order to properly make the case for opting in, and would involve considerable loss of face for NHS England. Even if a campaign for an opt in system succeeded on its own terms, it is highly likely that a much smaller overall percentage of the population would be enrolled than through an opt out system – reducing some of the potential advantages to be got through the system. So what is the best way forward from where we now are?
Getting the information provision right is a necessary first step, but is not a full solution. What will matter more over the long term is getting the governance processes right: creating a transparent and accountable system that allows patients to satisfy themselves that their trust in the care.data system has been well placed.
Choices about opt outs need to be presented to individuals in such a way that they are readily understood by those of normal intelligence and attention span. So there are limits to how useful information provision and opt-outs can be for individuals who are confronted with a large scale and complex project such as care.data. The more complex and fine-grained the choices are that individuals are asked to make, the greater the likelihood that the choices will be poorly understood.
The current system, with two separate opt outs, has proved somewhat confusing. But these two choices do not yet give many patients the control they would want. For example, the type one opt out (which prevents a patient’s data from leaving their GP’s system) does not allow patients to draw distinctions between different uses of their data. Many patients will be happy for their data to be used in NHS commissioning, or university research, but unwilling for it to be sold to commercial companies, leaving them in a quandary about whether it would be best to opt out or not. Moreover, the opt out at present applies to all data that would be extracted by the General Practice Extraction Service (GPES) – there is no separate care.data opt out.
What we require is a system that will allow care.data to make choices about uses of data that NHS patients are themselves comfortable with. I suggest that what is required is a robust governance framework that is based on the values that are already central to the NHS.
The fact that the NHS allocates treatments according to need and not ability to pay, funded on the basis of general taxation, has long been a source of national pride. We feel proud not just that common goods such as an integrated healthcare system are available free to everyone, but proud also that we mutually recognise the importance of these goods and are willing to play our part in sustaining them.
This willingness to do one’s bit for the sake of the common good is frequently presupposed by public health policies. For example, enough parents must ensure that their children get vaccinated if we are to gain the considerable benefits of herd protection against measles. With appropriate governance, care.data would fit neatly into this public health model of common acceptance of small individual costs for the sake of the common good. The risks of reidentification and breach of privacy could be rendered minimal (particularly if the system were changed to one in which the data was pseudonymised before it left GPs’ data systems), and the common good of public health can clearly be served by the system.
However, sustaining the solidarity and public trust necessary for this will depend on an honest dialogue about the size of risks are that individuals are being asked to consent to, and how the care.data system will best serve the public interest. There is something undeniably noble about voluntarily accepting minor risks for the sake of the common good, but being expected to accept such risks to enhance the profitability of corporations will for many be a step too far.
A lot is still to be decided during the pause. It is currently unclear what the principles are that will be used to determine who should be able to gain access to care.data and for what purposes, and it is also unclear which body will be making these decisions. The Health and Social Care Act 2012 requires only that the usage of data “be in the interests of the health service in England or of the recipients or providers of adult social care in England”. This purpose is broad, and poorly defined. It needs to be better specified through the core values of the NHS to ensure that care.data makes the contribution that it is capable of to the future of the NHS.
James Wilson is Lecturer in Philosophy and Health at UCL, where he is also the Director of the Centre for Philosophy, Justice and Health. His research focuses on ethical issues in the use of information, and in public health policy. He is the ethicist for the General Practice Extraction Service’s Independent Advisory Group. This article is written in a personal capacity, and should not be taken to represent the views of the General Practice Extraction Service, or the Independent Advisory Group.
The Royal Society, Science as an Open Enterprise. Makes the case for in an intelligent approach to open data – providing analysis of some of the privacy and economic questions.
The General Practice Extraction Service Independent Advisory Group, Public Interest Model. Provides a model for thinking about how the public interest should be determined in cases such as care.data. See also, Mark Taylor, “Information Governance as a Force for Good? Lessons to be Learnt from Care.data”, Scripted Volume 11, Issue 1, April 2014.
The MedConfidential website provides a blog of up to the minute developments, as well as instructions on opting out.