Research Exceptionalism

Abstract

Research involving human subjects is much more stringently regulated than many other nonresearch activities that appear to be at least as risky. A number of prominent figures now argue that research is overregulated. We argue that the reasons typically offered to justify the present system of research regulation fail to show that research should be subject to more stringent regulation than other equally risky activities. However, there are three often overlooked reasons for thinking that research should be treated as a special case. First, research typically involves the imposition of risk on people who do not benefit from this risk imposition. Second, research depends on public trust. Third, the complexity of the moral decision making required favors ethics committees as a regulative solution for research.

Keywords:

• ethics committees
• human subjects research
• professional ethics
• regulatory issues
• research ethics
• risk/benefit analysis

An earlier version of this paper was given at the International Congress of Bioethics in Rijeka. The authors would like to thank the audience there for helpful comments. James Wilson's work was undertaken at UCL/UCLH who received a proportion of funding from the Department of Health's NIHR Biomedical Research Centres funding scheme. All parts of the work were jointly written by both authors, and both authors contributed equally to the final manuscript.

Notes

In the United Kingdom, Australia, and New Zealand these committees are referred to as research ethics committees, while in the United States they are referred to as institutional review boards. We here use the terms interchangeably and primarily use either research ethics committee or REC to refer to whichever committee is responsible for the ethical review of research.

What specifically is legally required in terms of approval for research depends on jurisdiction. In almost all jurisdictions some or much of medical research is legally required to seek ethical approval. In some jurisdictions, such as the United States and Australia, university-based nonmedical research is legally required to seek ethical approval. In others, such as the United Kingdom and New Zealand, there is no formal legal requirement to seek ethical approval for university-based research unless it is medical research or research in an area that is covered by legislation (human tissue and research involving people lacking the mental capacity to consent). A relatively complete listing of legislation governing research can be found in Office for Human Research Protection (OHRP) (2009).

Hence our usage of the term is to be distinguished from that deployed in technical writing in risk, which distinguishes risk from hazard. In this literature, risk refers to cases where we can assign known probabilities to a given event—for example, a coin toss—whereas hazard refers to cases where we are unable to assign probabilities. Given that a good part of what makes research research is the fact that we lack knowledge about what the effects of our intervention will be, the regulation of research will tend to be focused more on hazard than risk. However, we have decided (as in common in philosophical discussions) nonetheless to prefer the term risk, but with the proviso that we shall use it in a broad way that also encompasses hazard.

There is a further complication here that we shall largely ignore: We have not specified to whom the event needs to be undesirable. For instance, no doubt tobacco executives all across the United States and Europe are currently working to avoid what they perceive as an undesirable event, namely, losses to their long-term profitability caused by the greater focus on tobacco as a public health issue. We would consider this to be a risk management problem, despite the fact that the measures to reduce smoking, which are undesirable from the tobacco executive's perspective, are presumably desirable from a more objective perspective. However, in this article we are interested in which risks we should be aiming to regulate as a society, so we focus only on events that would be undesirable from the perspective of a just society.

In the case of this study we might argue, for instance, that the quality of the care the patients receive would in some sense be compromised by the research process, or that the fact of their treatment being randomized without their consent wrongs them even if it does not harm them.

For example it has been estimated that it costs £800 for a research ethics committee to consider an application and £850 for a researcher to prepare it (Arshad and Arkwright 2008).

For instance, if we regulate to reduce speed in built up areas by installing a large number of road humps, this is likely to increase the risk that people being rushed to hospital for emergency treatment will die on the way.

This is because, as Wolff and de Shalit put it in a slightly different context, “If two goods, or two forms of advantage and disadvantage, cannot be compared, then they cannot be placed on a common scale, and so it will become impossible, in many cases, to say whether one person is worse off or better off than another” (2007, 23).

One obvious way is to do an expected utility calculation, by multiplying the probability of the bad event occurring, by the disutility of its occurrence, to give an expected utility score. However, it is far from clear that this is the best way to treat risk. As Wolff points out, it does not seem irrational for someone to be willing to suffer a 1 in 2 million chance of death for 1, but not to be willing to suffer a one in two chance of death for 1,000,000, let alone 2,000,000 for their certain death (2006, 61).

In addition, in real-world contexts we will typically be acting in the face of hazard, rather than risk, and so will not be in a position to judge accurately the probability either that the possible undesirable effects will obtain unless we act, or the likely effectiveness of our attempts to prevent the undesirable events. Hence even if we do think that we can provide an accurate index that will rank risks, given their probabilities, we may not be in a position to rank them on the basis of the information available to us.

In the United Kingdom the present National Health Service's Research Ethics Committee approval form is 34 pages long and must be accompanied by an information sheet, consent form, and a research protocol.

There are cases where a body of knowledge (such as body of knowledge that constitutes the science of torture) is so ethically problematic that it is wrong in itself to attempt to extend it. But very little, if any, of the research examined by ethics committees falls into this category.

Presumably because the information generated by an audit is not generalizable new knowledge but rather assessment of a local situation and whether it is meeting some preset standard. For further on the audit/research distinction, see Cave and Nichols (2008), Holm (2007), and Holm and Bortolotti (2007).

A further peculiarity of the current systems of research scrutiny is that projects may get described as research even if they do not fall under the definition just given; for example, student research projects even if they are effectively audits are often assessed by research ethics systems precisely because they are described as research projects. Likewise, the work of academics such as those in the visual or performing arts sometimes are scrutinized by research ethics committees despite there being no attempt to create generalizable knowledge as part of the process.

For instance, TV shows routinely reenact problematic pieces of research, such as Milgram's obedience to authority experiments, without seeking approval from an ethics committee before doing so. Note that these experiments would probably not be approved by any ethics committee now.

Presumably driving ought to be regulated fairly stringently, but there is no need for driving ethics committees.

For a concise overview of the history of research ethics, see Schüklenk (2005).

For some of the empirical evidence, see Featherstone and Donovan (2002). For philosophical reflection on what this should mean for our attempts to gain informed consent, see Dawson (2009).

See Hunter (2007).

See Anthanassoulis and Wilson (2009).

A fuller explanation of the experiments can be found in Milgram (1974).

See Elliott and Hunter (2008). It might be questioned whether research ethics committees are more likely to make ethically defensible judgments. However, it seems likely that they will be more reliable, given the plurality of different views represented on an ethics committee, the variety of experience and expertise in considering and identifying ethical issues in research, and general arguments from political philosophy such as the Condorcet Jury Theorem supporting the notion that groups will generally be more reliable decision makers than individuals (Hunter 2007).

It might be argued that professional bodies have codes of ethics that are a form of compromise position that may (if the profession knows what is good for it) represent the public's view. To some degree this is true; however, codes of ethics are blunt tools that require interpretation. The more complex the situation, the less use these will be in providing either reassurance or guidance. Given the complexity and variety of research, professional code of ethics will only go so far in terms of removing the need for regulation, and ethical review by committee will be needed.

One question we shall leave for further work is whether such arguments could justify the claim that research as such should be treated as exceptional. It might seem tempting to say that only certain categories of research (for instance, clinical trials) should be treated as exceptional, while other categories of research (such as questionnaires or interview studies) may legitimately be regulated in a less stringent fashion. However, there may be some quite sophisticated reasons for regulating all research stringently even though only some research raises sufficient moral problems to be worthy of stringent regulation. For instance, it might be the case that it is difficult to distinguish ethically problematic research from ethically unproblematic research, and that it is better, all things considered, to send all research protocols to a research ethics committee for scrutiny even though only some research protocols will be sufficiently ethically problematic to justify scrutiny by a full ethics committee. For an argument along these lines, see Hunter (2007).