More than any other group, the Learning Healthcare System will have implications for patients. It was strange therefore that it was patient representatives who were the most difficult group to recruit for interviews and seminars on this topic. Approaches to ten patient leaders or representative groups resulted in only one participant. While there were many logistical reasons why potential participants were unable to attend, this response rate was much lower than for any other group. This may reflect the general lack of awareness of the topic among the public (Cameron, Pope et al. 2014).
A large proportion of the population in most countries have experienced a huge increase in their use of computers, mobile phones and the Internet over the last 10 years. Among communication, shopping, banking, information retrieval, etc. health stands out as an area whose electronic interface with the public has remained relatively undeveloped. Healthcare is yet to go through the change that these other industries have experienced, where people have fully engaged with technology and online services.
There is a growing expectation for online interaction with the health service, as evidenced by the growth of services such as, Patient Opinion (Patient Opinion 2015) and Patientslikeme (Patientslikeme 2015). This follows a general trend whereby it is becoming more acceptable, at least for a proportion of the population, to share personal details online (Munro 2015). For others, they may feel comfortable booking appointments or ordering repeat medications, but it is unclear whether the majority are yet ready to use online services for sharing individual health information (Participant41 2015). It is also unclear what secondary uses of healthcare data the public will find acceptable.
A study in the UK identified some fear among the public that routinely collected health data might be used by Government to cut spending on healthcare or that sensitive information might fall into “the wrong hands”, with individuals being identified and stigmatised or disadvantaged in some way (Millican and Mansfield 2013). The same study found that health data is viewed differently from other types of data.
There is little objection to aggregated patient data being used for the general good, as long as commercial gain is not the priority. There is however, significant discomfort with the idea of health data being linked with non-health data, such as patient records with supermarket loyalty cards to inform health promotion messages. When part of a behaviour change intervention, this may even be viewed as a threat to free will. Other types of data linkage are less controversial (Millican and Mansfield 2013).
It was felt that, at least in the UK, there has been a loss of confidence in established authorities, with the proportion of people trusting the NHS to steward their data appropriately falling below 50% (Manning 2015).
Communication with patients and the public, about this process, is seen as critical. There have been examples when poor communication has actually increased worry about data sharing (Foley and Fairmichael 2015). There is a sense that the media is more sensitive to risks to confidentiality than to the harm that could be prevented by sharing data. Apart from confirming the adage that “bad news sells”, this phenomenon would appear to be an example of the behavioural economic effect that willingness-to-pay is usually lower than willingness to accept (McClelland and Schulze 1991). This is a major risk to the development of Learning healthcare Systems, but also has implications for how any argument for data sharing might be framed.
An informed public discussion is called for (Bates 2015) and in the US, the IoM is currently building a network of patient advisory councils around this topic (McGinnis 2015). Attitudes in the UK are likely to be quite different as there have not been the historical commercial interests in healthcare and the requirements to share data in order to get insurance coverage (Manning 2015).
Another UK study found significant resistance to the idea of linking routine data, among a sample of the public. However, when these people engaged in a two-day workshop on the topic, their views became significantly more positive (except in relation to commercial uses) (Cameron, Pope et al. 2014). This intensity of engagement is clearly not feasible on a population level, so thought must be given to how best to engage. Perhaps a strategy that focuses engagement on the harshest critics can help to develop a system that addresses most objections.
Another strategy is to target all public interfaces with the healthcare system, with a multimedia campaign, so that the concept of the Learning Healthcare System becomes common knowledge (Foley and Fairmichael 2015). Any campaign must also make it clear why we need a Learning Healthcare System. There may not be a realisation among a proportion of the public, that there are still huge gaps in our medical knowledge and in our ability to run safe, effective and efficient healthcare providers (Foley and Fairmichael 2015).
While recent controversies and their impact on programmes such as care.data might suggest a political push for ever-greater restrictions on the use of routine data, there is also a feeling that excessive controls will threaten the viability of Learning Healthcare Systems. For example, there is increasing use of secure data labs (Department of Health 2013). Traditionally, researchers might receive regular downloads of pseudonymised datasets, such as HES, that could potentially be maliciously re-identified. Increasingly, researchers will have to visit secure data facilities to do their analyses on computers with no access to the Internet. Results will have to be checked and approved prior to release from the facility. This approach has been shown to reduce public concern (Cameron, Pope et al. 2014), however, it could curtail many use cases of the Learning Healthcare System and would make international comparative work very difficult.
In clinical practice, the Learning Healthcare Systems offers the opportunity for clinicians to deliver truly patient centred care, by helping them to take account of patient preferences. For example, by allowing them to leverage many sources of information to help weigh investigation and treatment options according to what is important to the individual patient, rather than presenting a one-size-fits-all “best option”. This would require patients to move up the activation/engagement scale, from 1 towards 3, from those who do not want to understand or engage in their care, through to those who want to understand what is happening, but be told what they should do, to those who want to understand and manage their own care (Bates 2015).
Recent research has shown a positive link between activation level and other outcomes (Hibbard, Greene et al. 2015). It is important to consider the implications of this for the patient and their role in the system. Patients will need to be supported and educated to enable active engagement and the system must support this. Further research is required to understand how this can be achieved (Simpson 2015).
Information Governance (IG) is therefore critical to the viability of a Learning Healthcare System. Data must be, obtained, held, used and shared, within a robust, ethically based IG framework. In the UK, this issue was addressed by the Information Governance Review (2013) (Department of Health 2013).
The issue is complicated both by legality and by public perception. In the UK, an IG framework must take account of, case law, common law, several different statutes and decades of NHS policy statements. This appears to be a particularly complex situation compared to other countries. For example, in the US, the Health Information Portability and Accountability Act (HIPAA) is a single act of Congress that sets out what an organisation needs to do if they want to use patient data. Even the US system has been considered inhospitable to Learning Healthcare Systems (McGinnis 2015).
A New Ethical Framework
There is a feeling that, as Learning Healthcare Systems begin to blur the boundaries between clinical practice and research, it no longer makes sense to hold the traditional distinction between clinical (light touch) and research (onerous) ethical approval processes (Etheredge 2015, Faden and Kass 2015, Foley and Fairmichael 2015).
This view is supported by Faden et al. (Faden, Kass et al. 2013), who reject the idea that, from an ethics standpoint, clinical research and clinical practice are different. Their work represents the first attempt to create an ethics framework to support the transition to a Learning Healthcare System. They believe that there is a three-pronged moral justification for transitioning to a Learning Healthcare System:
• The establishment of a just healthcare system
• The achievement of high quality healthcare
• The achievement of economic well being
Their framework consists of seven obligations (Faden and Kass 2015):
1. Respect the rights and dignity of patients
2. Respect clinician judgments
3. Provide optimal clinical care to each patient
4. Avoid imposing nonclinical risks and burdens on patients
5. Address health inequalities
6. Conduct continuous learning activities that improve the quality of clinical care and health care systems
7. Contribute to the common purpose of improving the quality and value of clinical care and health care systems
These obligations fall, to a greater or lesser extent on, clinicians, researchers, healthcare managers, commissioners and providers. The seventh obligation falls on patients and the authors are clear that it cannot be discharged through financial payment alone, rather everyone must participate in learning activities, so long as those activities do not breach the first four obligations. This obligation on patients is based on what John Rawls calls the principle of the common good and is relatively new to healthcare (Faden, Kass et al. 2013).
This framework has the effect of making learning easier by reducing the overprotection of patients from learning activities that do not undermine their interests or rights. They argue that this over protection currently deters learning (e.g. through research), resulting in significant harm to patients. In practice, this could mean the use of a streamlined consent process or, in some cases, no consent process at all within a Learning Healthcare System (Faden, Beauchamp et al. 2014).
This framework challenges the relevance of traditional ethical frameworks, so it is interesting to note that one of its authors is Professor Beauchamp, familiar to all doctors as co-author of the four principles of biomedical ethics (Beauchamp and Childress 2001), the framework most commonly taught in medical schools.
This framework has had a mixed reception (Faden and Kass 2015):
- Significant enthusiasm from those supportive of learning health systems and from those in the bioethics community whose views are that multiple options for informed consent may be acceptable in different types of circumstances.
- More reservation amongst those doubtful about learning health systems in general and from those wedded to the view that the traditional informed consent model is the only morally acceptable approach for clinical research.
The idea of studies without informed consent has been controversial. This ignores the double standard that patients often don’t have informed choice in other situations, for example in quality improvement studies (Faden and Kass 2015). It should be acknowledged that there are alternatives to informed consent that are worth exploring. Engagement with patients and the public regarding which procedures need consent and which don’t, will be key to the development of this change (Faden and Kass 2015).
Most participants agreed that an opt out system was appropriate for the use of routine patient data within a Learning Healthcare System (Foley and Fairmichael 2015), however, it is also important to consider where the opt in or opt out sits in the process. For example, an opt in system might be appropriate if the decision is about whether commercial organisations should be allowed access to data. These decisions would depend on the particular use cases.
Implementing a New Ethical Framework
This framework could not be implemented immediately on top of our current healthcare system and would need a phased approach (Faden and Kass 2015). Some healthcare systems are currently more suited for this development than others. To be successful, like any kind of social change, implementation of the framework will require a number of steps (Faden and Kass 2015):
- The healthcare system leadership must have the vision and commitment to have an ethically robust LHS
- There should be a culture within the institution or system to endorse this approach.
- This may be easier in a system like NHS than in a fragmented system such as the US.
- Documents like the NHS constitution may be important and would be worth examining to determine whether it would need to be expanded or developed to facilitate this transformation
- Clinicians must be carefully involved
- Patients must be involved and play a central role in decisions about which types of studies or projects could go forward with which different types of consent or disclosure
The “ETA” approach is important in the development of the learning health system and is relevant to the ethical foundation of such a system (Faden and Kass 2015):
- Massive engagement of stakeholders, especially patients and the public, is required. It is important for them to understand the reasons for the system and that it will help to deliver current best care
- Patients should help decide what sort of studies should require full informed consent, streamlined consent or no consent at all
- Be open that your system is committed to learning from care
- Ensure that when joining a health system, the public is aware that the aim is to learn from care and to constantly improve
- Let people know when care does change as a result of learning activities. Use multiple methods including newsletters, websites etc.
- Accountability from the system side in making sure that care actually does change and improve as a result of learning. The justification for streamlined consent or disclosure approaches is that these will facilitate more learning, which is ultimately in patients’ best interests. If the learning does not translate into changes in care, then the ethical “contract” is no longer valid
The ETA approach was echoed by our focus group on Ethics and Information Governance (Foley and Fairmichael 2015).
Biobanks and Electronic Medical Records: Enabling Cost-Effective Research
Professor Faden and Professor Kass Interview
Dr Shaun O’Hanlon Interview
Mr Kingsley Manning interview
Dr James Munro Interview
Ethics and Information Governance Focus Group
Taking the Long View: How Well Do Patient Activation Scores Predict Outcomes Four Years Later?
Mr Lynn Etheredge Interview
Dr Lisa Simpson Interview
Dr Michael McGinnis Interview
Dr David W Bates Interview
Site visit to Geisinger Health System
Informed consent, comparative effectiveness, and learning health care
Patient and public views on electronic health records and their uses in the United kingdom: cross-sectional survey.
Ethics and Informed Consent for Comparative Effectiveness Research With Prospective Electronic Clinical Data
An Ethics Framework for a Learning Health Care System: A Departure from Traditional Research Ethics and Clinical Ethics
The Research-Treatment Distinction: A Problematic Approach for Determining Which Activities Should Have Ethical Oversight.
Ethical Oversight of Research on Patient Care
Conceptualising and creating a global learning health system
Research on Medical Records Without Informed Consent