Protecting the public from the adverse effects of confused research ethics

Neglecting treatment uncertainties leads to avoidable suffering

Healthcare abounds with uncertainties about the effects of treatments. When therapeutic uncertainties have not been addressed, subsequent research has gone on to show, in retrospect, that patients had suffered and died unnecessarily.^1,2 Suffering has occurred either because patients have been prescribed treatments that turned out to do more harm than good (steroids for traumatic brain injury, for example); or because helpful treatments had not been recognised earlier to be beneficial (for example, steroids given to women expected to deliver babies prematurely). In these ways, patients and the public have suffered from failures to facilitate and promote research to address uncertainties about the effects of treatment already being used in routine care.

Who is responsible for promoting the research needed to address treatment uncertainties and thus reduce harmful elements of healthcare? ³ A wide range of people could be considered to share in this collective responsibility, including research funders, researchers, health professionals and patient groups. A particularly heavy responsibility must inevitably lie with the health professionals who prescribe or withhold treatments with uncertain effects. This principle is made explicit in the section entitled ‘Maintaining and improving your performance’ in the UK General Medical Council’s 2006 expectations of doctors. Paragraph 14f in Good Medical Practice states: ‘You must work with colleagues and patients to maintain and improve the quality of your work and promote patient safety. In particular, you must help to resolve uncertainties about the effects of treatments’. ⁴

Although this expectation reflects a responsible professional standard, the General Medical Council unaccountably abandoned it in the 2013 edition of Good Medical Practice.^5,6

Obstacles facing proposed research to reduce treatment uncertainties

Several obstacles confront those healthcare professionals who wish to promote research to address uncertainties about the effects of their treatments. All of these obstacles can be reduced if there is a will to do so. ¹ For example, research can be commissioned to assess the uncertain effects of treatments that are of no commercial or academic interest, and incentives of various kinds can be used to promote collaboration in controlled trials of a size sufficient to assess treatment effects on outcomes that matter to patients. The RECOVERY trial, which randomised 20,000 patients to several different COVID-19 treatments, illustrates the potential ¹ : the National Institute for Health Research provided emergency funding to address priority questions and to support a trials network infrastructure. Significantly, the Chief Medical Officers of England, Scotland, Wales and Northern Ireland wrote to clinicians to urge them to treat patients within the RECOVERY trial platform rather than use medications ‘off-label’.

A further potentially remediable obstacle is hyper-regulation of research to compare treatments that have already been adopted in routine practice. ⁷ Since the introduction of research ethics committees (RECs) in the 1960s, surveys have shown repeatedly that the processes used and judgements made by RECs vary greatly.^8–13 When the same research protocol has been submitted to many RECs, committee judgements have ranged from very low adjudged risk and complete exemption from ethics review, to being so risky that the treatments should not be permitted. Fifteen years ago, we suggested in a BMJ editorial that ‘ethics review roulette’ is an intervention in the healthcare system that had been less evaluated than it should have been. ¹⁴

The obstacles presented by ethics review might be justified if it could be shown that, on balance, the burdens of the review do more good than harm to patients’ interests. That may well be plausible for treatment innovations, but seems very unlikely to apply to treatments that have already been adopted within routine care without sufficient assessment. Indeed, research hyper-regulation in these circumstances seems more likely to operate against the interests of patients and the public. Attempts by researchers to survey the functioning and effects of research ethics committees have met with only limited success. For example, in a recent study of how they function in the United States, only 40% of those approached agreed to participate. ¹⁵

After estimating the high costs of obtaining ethics and site-specific approvals for multi-centre research, Petrova and Barclay concluded: ‘If we continue to work under current systems, we are perpetuating, paradoxically, an unethical system of research approvals by virtue of its wastefulness and impoverished ethical debate’. ¹⁶

Adverse effects of ‘a confused ethical analysis’

Some obstacles are less tractable than others, particularly when they result from a confused ethics analysis that underlies an illogical double standard governing treatment. As an American medical sociologist once put it: ‘Doctors are at liberty to do anything thing they like in the clinic as long as they undertake not to learn from the experience!’

Richard Smithells, a British paediatrician, drew attention to the double standard many years ago when he observed: ‘I need permission to give a drug to half of my patients [to find out whether it does more good than harm], but not [if I want] to give it to them all’. ¹⁷

Two decades later, John Lantos, an American paediatrician and ethicist, pointed out that: ‘This confusing real world situation seems to reflect a confused ethical analysis’. ¹⁸

Seeds of this confusion were sown in the 1960s by commentators who did not distinguish clearly between research done to assess the effects of treatments already adopted in clinical practice and the many other types of research, some of which undoubtedly do need ethics assessment. In the UK, Maurice Pappworth’s 1967 book – Human guinea pigs: experimentation on man ¹⁹ – exposed some scandalous examples of unethical, mainly preclinical research, and the book was very influential in prompting the creation of research ethics committees. ²⁰

The title of Pappworth’s book raised the question of what was meant by the word ‘experimentation’. Illogically, and with no empirical evidence to support it, a damaging view became promoted and entrenched, namely, that the interests of the vast number of patients involved in the ‘uncontrolled experiments’ of everyday medical ‘tinkering’ are less in need of protection than are those of the much smaller number of patients who are involved in planned, properly controlled clinical experiments addressing demonstrable treatment uncertainties.^21,22 Pappworth failed to make this important comparison,^19,20,23 despite being challenged publicly to do so more than once.^24,25

How do professionals deal with uncertainties about treatment effects?

How do health professionals deal with uncertainties about the treatment options facing an individual patient with a chronic health problem when current treatments do not seem to be helping? An informal ‘try-something-else-and-see’ approach is a common way of comparing alternative treatment options: ‘If Treatment A doesn’t seem to be helping, give Treatment B a try’. However, these informal approaches can yield untrustworthy evidence because biases and the play of chance have not been adequately considered.

Ideally, faced with such treatment uncertainties, the treatment options should be compared fairly after randomly allocating the order in which the treatments are given to the patient – a so-called randomised ‘n-of-1’ trial, in effect, an evidence-based form of personalised medicine. However, quite apart from the lack of infrastructure to facilitate use of such trials in routine care,^26,27 proposals for randomised n-of-1 trials can become victims of the double standard. This is because many ethicists and other research regulators believe that patients whose treatments are chosen using random allocation to inform decisions about their care are in greater need of protection than those involved in the informal ‘try-it-and-see’ approaches to choosing the same treatments on hunch or some other basis in everyday care (see fourth video clip at https://healthtalk.org/clinical-trials/amanda-interview-22#ixzz39vlvFlb). In a recent survey of members of Australian research ethics committees, for example, more than two-thirds of the respondents judged ethics approval to be required for randomised n-of-1 trials. ²⁸

Even if randomised n-of-1 trials were used more widely to generate fair comparisons of treatment periods ‘within’ individual patients, they can only be employed in a minority of treatment circumstances, usually chronic health problems. More usually, when there is uncertainty about the relative merits of different treatments, fair comparisons of alternative treatments are made after random allocation has been used to assign patients to one of two or more treatment comparison groups. In these circumstances too, however, ethicists and other research regulators believe that patients are in greater need of protection than those prescribed the same treatments in usual care.

As any behavioural scientist knows, creating obstacles decreases uptake. The impact of obstacles was illustrated by a demonstration project involving the use of routinely collected electronic health records to run pragmatic controlled trials comparing widely used elements of healthcare – antibiotics for acute exacerbations of chronic obstructive lung disease and statins to reduce cardiovascular risk. ²⁹ The research governance approval process for studying these widely used interventions took over three years. Of 459 English and Scottish general practices contributing electronic health records to a research database, 270 expressed interest in participating, but only 17 (6%) went on to participate in the trials. Governance and consent procedures substantially compromised the intended simplicity and successful conclusion of the trials. For example, Good Clinical Practice guidelines (designed to govern the prescription of novel drugs) required participating clinicians to be formally ‘trained’ in prescribing treatments they had been prescribing for decades! The report of the demonstration project concluded that:

Electronic health record point-of-care trials are feasible, although the recruitment of clinicians is a major challenge owing to the complexity of trial approvals. These trials will provide substantial evidence on clinical effectiveness only if trial interventions and participating clinicians and patients are typical of usual clinical care and trials are simple to initiate and conduct. [Our] recommendations for research include the development of evidence and implementation of risk proportionality in trial governance and conduct. ²⁹

Improving the relevance and efficiency of research ethics review

Research ethics committees have sometimes helped to protect the interests of patients and the public, but they have often failed to do so. We drew attention to some of these failings in 1995 (Savulescu et al. ³⁰ and Box 1a) and warned patients and the public about them in 2009 ³¹ and 2011 (Evans et al. ³² and Box 1b). We noted in 2014 that RECs had failed to ensure that proposals for treatment research had taken account of systematic reviews of what was already known (Chalmers et al. ³³ ; Box 1a: item 1; Box 1b: item 2). We also observed that RECs had not used their powers to reduce biased under-reporting of the research that they had approved ³⁴ (Box 1a: items 3–5; Box 1b: items 1 and 3).

Box 1.

Summary of recommendations.

(a) Research ethics committees should: (1) Require systematic reviews of existing research before approving research (2) Require that a summary of relevant systematic reviews be made available to potential participants (3) Require registration of clinical trials at inception as a condition of approval (4) Require a commitment by investigators to make the results publicly accessible as condition of approval (5) Audit the reporting of results of research previously approved by them.

(b) People invited to participate in controlled trials should agree, on condition that: (1) the study protocol has been registered and made publicly available (2) the protocol refers to systematic reviews of existing evidence showing that the trial is justified (3) you receive a written assurance that the full study results will be published and sent to all participants who indicate they wish to receive them.

The relevance of research ethics committee assessments could be improved dramatically by ensuring that longstanding recommendations are addressed. As Clarke ³⁵ suggests in his contribution to Principles of Health Care Ethics, ³⁶ for example, ‘by not doing a systematic review before embarking on their trial and by failing to report new results in the context of an updated systematic review, researchers are failing both ethically and scientifically to conduct the best possible research’.

The requirements of ethics committees and other research regulators can delay or inhibit research beneficial to patients, ²⁸ and they have even sometimes led to death resulting from delays in initiating treatment. ³⁷ Importantly, hyper-regulation of research, and the resulting delay in identifying beneficial and harmful effects of treatments, has adverse consequences for all patients, not just the participants in research seen through the current restricted focus of ethics committees and other research regulators.^2,21

Because acquiescence in uncertainties about the effects of treatments given in usual clinical practice can result in avoidable harm, promotion of controlled trials within routine care should be an expected element of professionally responsible care ³⁸ and promoted by research ethicists. This principle has been reflected in the clinical practice and research of the paediatric oncology community and in a steadily improving prognosis for children with some haematological malignancies. ³⁹ As Richard Ashcroft, a British medical ethicist, has suggested, for ethical as well as scientific reasons, when there is properly informed uncertainty about the relative merits of alternative treatments, ‘the trial is the treatment’. ⁴⁰

Reducing the adverse effects of confused medical ethics

The price being paid by patients and the public for acquiescence in treatment uncertainties is inadequate prevention and relief of health problems. In Testing Treatments (which is freely available in 20 languages at www.testingtreatments.org), we have urged readers to ‘Encourage and work with health professionals, researchers, research funders and others who are trying to promote research addressing inadequately answered questions about the effects of treatment which you regard as important’. ³²

It is tempting to assume that those who are made aware of the problems we have described would agree that – emulating the processes developed in paediatric oncology to help children with cancer – treatment uncertainties should be addressed using randomised treatment comparisons embedded within routine care. However, Michelle Meyer et al. have found that people often approve of universal implementation of untested policies or treatments while disapproving randomised experiments to assess which of the policies or treatments is superior. ⁴¹ These findings indicate a need for research to help increase an understanding of what people know and believe about the consequences of acquiescence in treatment uncertainties; what they think should be done about them and why random allocation to treatment comparisons is so widely promoted as a feature of the research required. ⁴²

We have noted how confused research ethics can and has compromised the best interests of patients and the public, sometimes lethally. Like other professionals working in or for the health services, research ethics committees and other research regulators should be held to account for the broader consequences of their policies and decisions, not only for their handling of individual study proposals.

In a section of the magisterial, 800-page book Principles of Health Care Ethics, the British medical ethicist Richard Ashcroft considered how the performance of research ethics committees might be evaluated. ⁴³ He suggests considering these committees as ‘public health interventions acting … to reduce, control or allocate the exposure of the population to the relative risks of research participation and, by the same token, the exposure to the relative benefits’ (our emphasis). Ashcroft goes on to develop this important consideration as an approach to the moral and practical significance of research ethics committee variation:

From a regulatory point of view, due process and fair consideration take very high priority. Hence one of the key challenges is to ensure that the process of review is carried out appropriately with reference to the norms of procedural justice. Amongst other things, this includes a right of appeal, reference to public and accessible standards, reasons given for decisions, and decision within a reasonable time. Of course, justice also requires that the participants be treated fairly. But it might require that non-participants and future patients be not unfairly excluded or delayed from accessing the benefits of research, a burden of responsibility which might arguably fall on the REC as much as on the researcher or sponsor (our emphasis). (Ashcroft, ³⁶ p.686)

Neglect of important uncertainties about inadequately assessed treatments used in everyday clinical practice poses a threat to health. A confused ethical analysis continues to frustrate efforts to reduce this threat.^44–46 We have attempted to show why the responsibilities of RECs extend beyond judging individual research proposals. The British ethicist Richard Ashcroft has made the welcome suggestion that RECs might take into account and use their influence to promote and protect the interests of non-participants in research and future patients.

Effective leadership within the research ethics community will be needed to challenge illogical and deeply entrenched current thinking and practices. The community must be more ready than it has been to welcome reasonable criticism of the research governance obstacles that stand in the way of addressing treatment uncertainties more effectively. Over the 17 years since we first broached this topic together, ¹⁴ there have been some encouraging examples of willingness to tackle entrenched attitudes and policies. In an accompanying article, ¹ we use some of these examples to illustrate the practical steps that can be taken to tackle these longstanding unethical and costly problems.

Acknowledgements

We are grateful to Amanda Burls, Hugh Davies, Elina Hemminki, Robert Matthews, Anna Mae Scott, Matt Westmore, Dominic Wilkinson, and the anonymous reviewers, for providing relevant information and/or commenting/help with previous drafts of this text.