Viewpoint: Why our conceptions of research and practice may not serve the best interest of patients and subjects

The distinction between clinical research and medical practice dominates in critical respects how we think about the landscape of biomedical ethics. It guides basic distinctions between medical ethics and research ethics, and it serves to bifurcate the biomedical world into two separate domains governed by separate systems of moral norms: the domain of accepted practice and the domain of research. But do we have good reasons to warrant a very sharp distinction between research and practice? Does this bifurcation need to be reconceived so that it does not ill-serve the interests of patients and subjects? Perhaps patients are today underprotected by our moral arrangements, and research subjects overprotected. Often we have worried that we underprotect human subjects. Much concern has centred on vulnerable groups, such as children and the mentally handicapped. These problems are prominent parts of the landscape of research ethics. But other problems are less explored features of our system. For example, the ethical review of research protocols is generally performed within and by employees of the very institutions in which the research will be carried out, a clear problem of conflict of interest. This problem of conflict of interest is a cornerstone of the system of research review, and this system of review clearly does not always function to protect the best interests of research subjects. Other problems in our system have not to do with the underprotection of subjects, but rather with overprotection. In some cases, potential subjects of research are more interested in being included in the research than they are about risks of harm. They do not see research as burdensome, but as potentially beneficial – and indeed, as holding therapeutic promise. AIDS activists, amongst others, have pointed out that – since the 1970s – law and public policies have often been so protectionist of human subjects that they fail to provide benefits to a segment of the population deeply in need of those benefits. This circumstance certainly looks like a moral failure. AIDS activists sought justice and respect for their autonomy and rather less in the way of protection against risk. To them, protectionist rules of research ethics excluded them from the benefits of research, including access to potentially beneficial drugs. Significant parts of government regulatory systems that use a sharp research-practice distinction date from the 1970s, many of them deriving from the conclusions reached by the US National Commission for the Protection of Human Subjects of Biomedical and Behavioural Research [1]. This Commission advanced the view that research protocols involving human subjects should have a high level of peer review, a level not required for patients in medical practice. The distinction between research and practice was at the time of this Commission blurry because research was often designed to evaluate and upgrade therapies, leading to the then widely used, but soon to be questioned, category of ‘therapeutic research.’ The line between research and practice was also obscured by departures from standard practice said to be ‘experimental,’‘innovative’ or ‘nonvalidated,’ without clearly distinguishing a research dimension from a treatment dimension. The National Commission determined that these terms and distinctions had never been carefully studied or defined, and the Commission set out to reverse this trend and to recommend new policies of the oversight of research. The National Commission found that, ‘For the most part, the term “practice” refers to interventions that are designed solely to enhance the well-being of an individual patient or client and that have a reasonable expectation of success.’ The Commission formulated the meaning of the contrast term, ‘research’ as follows: It ‘designates an activity designed to test a hypothesis, permit conclusions to be drawn, and thereby to develop or contribute to generalizable knowledge’ [1]. The Commission’s view can be bluntly stated: To qualify as [medical] practice, two conditions must be satisfied: (i) an intervention aimed at benefiting a patient and (ii) the intervention has a reasonable probability of success. By contrast, to qualify as research, the two conditions that must be satisfied are (i) protocol-controlled design to test a hypothesis and (ii) an organized aim at generalizable scientific knowledge. This elegantly simple schema has had enormous influence in biomedical ethics and regulatory policy internationally, and it appears still today to be the received view. Because of a history of abuses of subjects in biomedical research, there arose in the 1970s and later years a prevailing assumption that research using subjects is dangerous, exists for the benefit of future patients and is aimed at scientific knowledge, not benefit for immediate subjects. In the Code of Federal Regulations in the United States, the term ‘research’ means ‘a systematic investigation, including research development, testing and evaluation, designed to develop or contribute to generalizable knowledge’ [2]. This conception still today is the received view. However, there are many problems with this definition. First, it uses the notion of research to define the term ‘research,’ creating problems of circular definition. Second, it does not define any of the several important terms (the key conceptual conditions) used in the definition, such as ‘systematic investigation,’‘testing’ and ‘generalizable knowledge,’ and these terms can be understood in several ways. Third, the definition is vague and overly broad because it is not clearly confined to biomedical research, clinical research and behavioural research – or even to scientific research, more generally. Its scope is left unclear. Fourth, and perhaps most importantly, it does not preclude ‘research’ from having a very close tie to ‘practice.’ Because the definition is so nonspecific, regulatory requirements that use the definition may judge that some activities that are questionably research involving human subjects nonetheless must be treated as such. Government requirements are today commonly applied even if ‘human subjects’ may not need to be protected by the rules of human-subjects research. A sweeping – that is, all-inclusive conception – of ‘human-subjects research’ can have immediate and unjustifiable practical impact on attempts to upgrade medical care, as I will now illustrate by use of a recent case. Dangers to patients and to health care institutions are dramatically present in this case [3–5]. It involves alleged human-subjects research on catheter-related bloodstream infections, which cause thousands of deaths each year in intensive care units (ICUs) [6]. In this case, regulations intended to protect research subjects wound up delaying preventive measures and benefits in hospitals for patients and could have eventuated in penalties to the medical research institutions and hospitals involved. This case has been reported in both the New York Times and the New England Journal of Medicine, but has not received as much attention as it merits under the topic of this conference, viz. the dangers of protective conceptions in biomedical ethics and regulatory schemes that might delay such work from being institutionalized in medical practice. I will also be arguing, using this case as an entree, that our system for the protection of human subjects of research sometimes reveals a crippling vagueness as well as a lack of good moral and conceptual judgment at the highest levels of protecting subjects from risk. In this case, Dr Peter Pronovost at the Johns Hopkins University was working with 103 ICUs in 67 Michigan hospitals to implement and evaluate, what had been established at Johns Hopkins and other intensive care units to be a very successful infection-control measure. However, the work was halted by federal regulators in the Office for Human Research Protections (OHRP) – the US’s highest office for the protection of human subjects. The allegation was that Pronovost and the 67 Michigan hospitals were using patients in human-subjects research without informed consent – obviously a serious allegation of an ethical violation. Pronovost’s funded activities were part of a study to improve medical care; it was sponsored by the Michigan Hospital Association. The aim was to improve care in ICUs by strictly implementing preventive procedures that had already been recommended by the Centres for Disease Control and Prevention, such as washing hands, using certain infection-control precautions, and the like. The team studied the effect on infection rates of a careful implementation in practice of all the recommended procedures following a checklist. They found that infection rates fell substantially if the checklist is scrupulously followed. After a published report of the study, the OHRP received a complaint that the research violated US regulations. After investigating, the OHRP demanded that Johns Hopkins (and Michigan hospitals) correct this mistake and undertake full review of the study. The Johns Hopkins IRB, an experienced review committee, had already examined the project and found that full IRB review and informed consent were not required for such a study. Here, trouble started. This IRB evidently had a very different understanding of federal regulations than did OHRP – a telling sign that regulatory statements are vague and unspecified, whilst assumed to be perfectly clear. One problem is the great unclarity that surrounds that concept of ‘research involving human subjects.’ If an IRB has one interpretation, and OHRP another, both research and practice can be held up and might even lead to disastrous federal penalties if the wrong judgment is made. The Pronovost case is stunning in this regard, because the activities produced no risk whatever for patients, and it was a way of following the safest practices known to exist – without introducing any research intervention with patients. These practices had already been approved by another federal agency or medical practice organization (and already proven safe and efficacious). The procedures used actually reduced risks for patients. The OHRP made the wooden judgment that because infection rates were being studied in patients, the work called for full committee review, for the informed consent of ‘subjects’ and for the whole apparatus of federal requirements for research involving human subjects. But Pronovost’s work was basically an attempt to improve medical care for patients without introducing the risks involved in doing research involving human subjects. There were no new interventions with patients. There were not even subjects in any meaningful sense. In my view, such work should not need an institutional review board approval even as expedited review. This is not research involving human subjects who are patients. In this case, regulations intended to protect research subjects caused a delay in the use of effective preventive measures in hospitals and could well have eventuated in unnecessary and foolish penalties to the medical research institutions and hospitals involved. It might also have caused unnecessary deaths of patients as a result of the delay. Somewhat happily, on 15 February 2008, OHRP issued a statement that in effect admitted that it had been wrong. It acknowledged that the work is ‘being used solely for clinical purposes, not medical research or experimentation.’ OHRP acknowledged that the activity, from the start, ‘would likely have been eligible for both expedited IRB review and a waiver of the informed consent requirement’ [7]. This confession of error is an honest acknowledgment, but it does little to restore faith in the system that led to the problem in the first place. Although government regulations always need some form of interpretation, we should not have a system in which thousands of lives might be lost because we have an outmoded conception of human-subjects research that leaves little room for riskless studies aimed at improving medical practice. Moreover, we want our medical systems to be improved by precisely this kind of study of practice – not to be retarded by the ethics of regulations. In Pronovost’s research, the system worked more to damage than to protect patients. One might object to my thesis using the argument that Pronovost’s research was not particularly deep or good because it falls far short of a randomized clinical trial. However, I reject this thesis. No doubt RCTs are the gold standard in biomedical research, but this fact does not mean that other forms of research or study are not of high value. Pronovost’s study is elegantly simple, but also clinically of the highest importance. US regulations also currently reject a once influential distinction between therapeutic and nontherapeutic research. This distinction was basic at the time of the National Commission in the framework of the Declaration of Helsinki, which made a pointed contrast between ‘clinical research’ that is merged with professional medical care and ‘nontherapeutic biomedical research’ that is not so merged [8]. The National Commission found this distinction obfuscatory and morally dangerous because it allowed research having only one or two therapeutic components – but otherwise was nontherapeutic – to be classified as therapeutic and clinical and therefore to be justified in the way other therapies were justified. Robert Levine, who has closely followed the National Commission’s views, has called this moral vision ‘the fallacy of the package deal’ [9]. The fallacy, he says, lies in the way nontherapeutic components of research protocols might be packaged with therapeutic components, and then the entire package ‘justified’ on grounds that a therapy is being offered. The distinction between therapeutic and nontherapeutic research has a considerably diminished presence in biomedical ethics today. No doubt we need to be cautious about this distinction, but I am not convinced that the language of ‘therapeutic research’ is illogical or a poor way to understand how research functions in many settings. Instead of abandoning the therapeutic-nontherapeutic research distinction, we might consider resuscitating it for today’s medical environment. Alexander Capron pointed out a quarter of a century ago that the term ‘therapeutic research’ makes good sense if it is defined as follows: ‘Research to evaluate practices or procedures that are intended to provide, or that have some reasonable possibility of providing, therapeutic, diagnostic, or preventive health benefits to [the] subjects’ [10]. So understood, ‘therapeutic research’ brings therapy and research together and focuses them on the individual patient-subjects involved in the research. We might today assess our situation in medical and research ethics as standing in need of a reversal of Levine’s idea of the fallacy of the package deal: The more research focuses on and stands to benefit immediate patients and shows promise of this benefit for them, the more the label ‘therapeutic research’ makes sense; but the more research is remote from a reasonable prospect of benefit for immediate patients, the less there is warrant for the label ‘therapeutic.’ The Pronovost study of Michigan hospitals falls well into the therapeutic domain. Particularly interesting is that, though the work is research, it does not appear to involve research involving human subjects who are patients. That is, it should not be placed in the category of ‘research involving human subjects.’ Clinical research today is often closely in contact with clinical medicine. In this environment, the research-practice distinction, and our understanding of clinical ethics and research ethics, may need increased flexibility and modification. As the examples I have mentioned suggest, we may already be behind in our thinking about how to make the needed adjustments to best protect the interests of patients and subjects. No conflict of interest was declared.