Background

During the COVID-19 pandemic many countries made very similar changes to research ethics and to approval processes for pharmaceutical products and medical devices.Footnote ³ In relation to processing applications in the research ethics and regulatory systems there was a strong focus on increasing speed and efficiency, so that COVID-19-related projects and products could quickly be assessed and approved. This involved changes, including (1) general streamlining of approval processes, (2) relaxation of some process requirements, (3) COVID-19-specific fast tracks,Footnote ⁴ and (4) a pause in the approval of research not related to COVID-19 to make space for COVID-19-related research. These were changes to the research ethics process, but not to the principles and rules by which research ethics bodies made their decisions.

Many countries also relaxed some substantive research ethics rules. These changes ranged from relatively trivial changes such as broadening the criteria for accepting information given orally and oral consent in clinical research to more substantial changes such as relaxing the rules for access without consent to health and other data.Footnote ⁵

Some scholars in the philosophical and bioethical literature argued that the COVID-19 pandemic justified further and more radical changes to research ethics rules and principles than those that were implemented in most jurisdictions. The fundamental structure of these arguments is remarkably similar across different authors and can be summarized in the following schematic argument:

Versions of this schematic argument provide the justification for many of the changes to research ethics proposed during the COVID-19 pandemic, although often with a number of the premises not fully stated. This is probably most fully stated in the academic literature arguing for the permissibility of COVID-19 human challenge studiesFootnote ⁶ and by the advocacy group 1Day Sooner in relation to such studies.Footnote ⁷ In the argument the phrase “the COVID-19 pandemic” can without loss of validity be substituted in the premises in which it occurs by any other situation that can be described as “a major, urgent, public health crisis.” This line of argument is often combined with an argument that the public health crisis created by COVID-19 is exceptional and that an exceptional crisis justifies exceptional measures.

Very similar arguments were made in relation to the methodological quality of research during the COVID-19 crisis where some argued for a relaxation of methodological standards in order to speed up research and get evidence faster. These arguments have been heavily criticizedFootnote ⁸ and are outside the scope of this essay. It is, however, worth noting in passing that insofar as methodological exceptionalism was actually implemented, it led to a crisis in evidence production and an epidemic of retractions of preprints and published articles.Footnote ⁹

Changes to research ethics during the COVID-19 pandemic

This type of argument has been used to argue for more specific changes to research ethics, ranging from the kind of process changes implemented in many jurisdictions to significant substantive changes to research ethics principles. It is impossible to list all of the suggested substantive changes, but they include conscription to vaccine research,Footnote ¹⁰ allowing research that puts research participants at significant risk of injury or death,Footnote ¹¹ allowing researchers to pay participants to take on significant research risk,Footnote ¹² and so on. There are specific counterarguments against many of the proposals for changes in research ethics principles, such as those in the now extensive literature on COVID-19 human challenge studies.Footnote ¹³ In this essay we are, however, focusing on the underlying, common, in-principle argument outlined above.

There are many other examples of suggestions for changes in research ethics that some claim are justified by the COVID-19 pandemic. The apogee of this line of argument was probably reached when some claimed in all sincerity that COVID-19 provided an urgent reason for the release and use of data and results from the infamous medical experiments conducted by Unit 731 of the Imperial Japanese army in Manchuria prior to and during World War II.Footnote ¹⁴ The relevant microbiological research performed by Unit 731 was on possible bacterial biological weapons and (obviously) done with methods available in the 1930s and early 1940s and on the basis of biological knowledge that was available at that time. It is not logically impossible that the research could be of value in relation to COVID-19, but it is exceedingly unlikely. COVID-19 is a viral disease and interacts with the body in fundamentally different ways than bacterial diseases do. Whatever knowledge Unit 731 developed about bacterial diseases would be based on what today would be seen as exceedingly simple, obsolete, or false biological theories. There are good reasons for demanding the release of all Unit 731 data from Japanese, American, and Russian archives to enable a full understanding of the atrocities committed, but the COVID-19 pandemic adds nothing to these reasons.

The basic principles of research ethics and the general structure of the research ethics system are both the results of prior crisis moments in medical research.Footnote ¹⁵ The primacy of the interests of the participant, the requirement for consent, and the right to withdraw from research as stated in the Nuremberg Code was a direct response to the research atrocities perpetrated by German and other researchers in the concentration camps during World War II.Footnote ¹⁶ World War II and the Cold War that followed it were seen as existential crises in many democratic countries; this perception of crisis was used as an argument for performing potentially harmful biomedical research without consent and sometimes without even the knowledge of the research participants.Footnote ¹⁷ When the Nuremberg Code was “rediscovered” in the 1960s, it became evident that the principled protection it put in place for research participants was relevant not only to Nazi doctors, but also to much research taking place in the U.S. and the U.K.

In a similar way, the general structure of the research ethics system as a regulatory system providing “event licenses” to researchers to conduct a specific project was a response to the increasing realization in the late 1960s and early 1970s that medical research was still taking place in ways that were deeply unethical and, in many cases, contravened the Nuremberg Code. It was primarily because of the spotlight put on highly problematic medical research by doctors such as Henry BeecherFootnote ¹⁸ and Maurice PappworthFootnote ¹⁹ and through the shock of public revelations of the Tuskegee study that we ended up with a regulatory system based on the prior approval of research protocols by an independent committee with lay representation.Footnote ²⁰ This particular crisis showed convincingly that research ethics could not be left to researchers and health-care professionals. Research ethics is thus born out of crises and as a response to what researchers are seemingly willing to do and policymakers seemingly willing to allow during perceived crises.Footnote ²¹ Over time, research ethics has been codified in a number of international normative documents with varying legal status.Footnote ²²

Justifying research ethics exceptionalism: Social value

Is the schematic argument outlined above valid and sound, leading to a true conclusion that research ethics must be changed in a major public health crisis if there is at least one possible intervention X that makes P5 true and one or more of P6–P6** also hold?

The first observation to make is that P5 as stated so far is ambiguous between “can definitely” and “can potentially.” In many cases, we do not yet know whether a particular intervention X is effective or not and the main point of researching X is to generate evidence about the effectiveness of X. So, in most cases, P5 should be stated as P5*: “X is a medical intervention that can potentially reduce the net negative consequences of the COVID-19 pandemic.” There may be cases where we already know that X is effective and where the research needed is therefore about, for instance, how X can best be deployed. In those cases, P5 will hold in its original form. It is, however, much less likely that P6 or P6** will be true if an intervention is already known to be effective. Thus, when P5 is true and can be read as “X … can definitely reduce,” then the relevant P6 premise will almost always be P6* and a procedural rather than a substantive change to research ethics would need justification.

In cases where P5* is the relevant premise because it is possible that X is effective, a question arises about the importance of the ex ante likelihood of X being effective to the validity of the argument from P1–P5* to IC1. Is it true that “Research into X has large social value and is urgent,” if the ex ante likelihood of X being effective is small? Seen in isolation and focusing on only one intervention at a time, the answer to this question depends on whether or not you believe that von Neumann-Morgenstern expected utility theoryFootnote ²³ as a model for rational choice holds in situations where an outcome has a small likelihood but a high positive utility if it happens. To put it slightly more contentiously, it depends on whether or not you believe that the consequentialist is pulling a “trick” when claiming that the possibility that X could save millions of lives is a convincing and dispositive reason to making the pursuit of X the overriding goal, even if the ex ante likelihood is low of X actually saving millions of lives. In an article arguing for the acceptability of human challenge studies and a range of other controversial COVID-19 interventions, Richard Chappell presents the consequentialist argument with admirable clarity in relation to the induction of early targeted immunity by deliberate viral inoculation, “variolation” of healthy volunteers. (He claims, somewhat unconvincingly, elsewhere in the article that this type of argument is not consequentialist.) Chappell argues:

In this quotation something admitted to be “obviously all speculative” becomes “entirely possible,” “prima facie plausible,” and then finally “a reasonable possibility,” leading to the conclusion that decision-makers “irresponsibly condemn[ed] to death many thousands who very well might have been saved.” However, because the whole argument is truly (as admitted explicitly by the author) “obviously all speculative,” no one had condemned anyone to death. There are no people who “very well might have been saved,” because the “very well might” only comes about through a series of unargued-for elisions from mere logical possibility to “reasonable possibility.”

In this way, even a minimal likelihood of an intervention actually working can be converted to a large, expected benefit and an imperative to act, if the projected public health crisis is large enough. In many ways, this argumentative form is the reverse of the (in)famous slippery-slope argument. Instead of automatically sliding down the slope, we are effortlessly carried up the escalator by the promise of a pot of preventative or therapeutic gold in the future.Footnote ²⁵

The ex ante likelihood of whether an intervention is effective, though, is important in order to estimate the moral weight it creates; it will not do to say that in a public health crisis any likelihood is good enough. This can perhaps most easily be seen if we consider the fact that resources for research are not infinite. Resources are always limited, even during periods where governments open their purse strings because of a crisis. This is true of financial resources, such as research funding, but it is also true of experienced researcher time, human participants for participant recruitment, and many other nonfinancial resources (some of which are nonfungible). This means that as soon as there is more than one intervention to research and there is competition for any of the necessary finite resources, a decision will have to be made about how to allocate the available resources. In that decision it will be crucial to estimate the likely effectiveness of each intervention as well as the likely magnitude of their effects. If we cannot estimate those two ex ante likelihoods for each of the interventions, we have no basis to make a rational choice or any choice at all about the allocation of research resources. Perhaps more importantly, in relation to the argument that in a public health crisis a minimal likelihood of effectiveness is sufficient to make an intervention something we should pursue in research, comparative analysis shows that this is not true unless pursuing the intervention that has ex ante minimal likelihood of effectiveness is costless.

We also need to note that predictions about the likelihood of research leading to effective interventions is notoriously uncertain. One example related to a still ongoing major epidemic illustrates this. It was conclusively shown in 1983–1984 that Human Immunodeficiency Virus (HIV) is the causative agent of Acquired Immunodeficiency Syndrome (AIDS) and it was predicted in 1984 that a vaccine would be available for testing in two to three years.Footnote ²⁶ Such a vaccine would obviously have great social value and there is consensus that the AIDS crisis can only be completely overcome when an effective vaccine becomes available.Footnote ²⁷ There has been significant investment and recent progress in HIV vaccine research over the years,Footnote ²⁸ but almost forty years after the prediction of two to three years, we still do not have an effective vaccine.

Furthermore, because what drives the argument is the large social value of having effective responses to a public health crisis, we also need to consider the likelihood that an intervention X that has been shown to be effective is actually implemented as well as the speed of implementation or gap between evidence generation and implementation. In many cases, the knowledge that something works will not lead to its immediate implementation and, unless the intervention is actually implemented, the large social value is not generated.

In many cases, the main delaying step between X being hypothesized to be a possible intervention on the basis of some initial evidence and the implementation of X is not the research and evidence-generating step. The main delay often occurs in the steps after this, for example, scaling-up production if X is a pharmaceutical product or the practical introduction of X in the health-care system or other social systems. If the implementation of X is global, the main issue delaying implementation is almost always global inequality. The effects of global inequality were also evident during the COVID-19 pandemic. Citizens in high-income countries (HICs) were vaccinated long before citizens in low- and middle-income countries (LMICs); at the time of writing, there are still many countries where less than 70 percent of the eligible population has been vaccinated.Footnote ²⁹ This may in some cases be due to vaccine hesitancy, but in others the main cause is unavailability. There is thus a great gap between the possible global social benefit of the development of a COVID-19 vaccine and the actual realized benefit. Such gaps between possible benefits of developing an effective intervention and actually realized benefits are completely predictable. For instance, even though an equitable global distribution of COVID-19 vaccines would have been highly preferable from the point of view of global justice and from the point of view of effective suppression of the pandemic,Footnote ³⁰ it was completely predictable that the distribution would be highly inequitable and that vaccination in LMICs would be significantly delayed. The policy imperative for policymakers in HICs of getting their own populations vaccinated as quickly as possible was always likely to outweigh more abstract considerations of and commitments to global justice.Footnote ³¹

In relation to estimating the social value created by research into a particular intervention there is also a potential issue of preemption that can occur, even if the intervention is quite likely to be effective. This issue is clearly illustrated by the development and deployment of COVID-19 vaccines, but it has also occurred in some other areas of COVID-19 therapeutics. Let us briefly look at the development and use of COVID-19 vaccines. As early as April 2020, 115 vaccines were in development following the publication in January 2020 of the genetic sequence of the original Wuhan strain of the SARS-CoV-2 virus.Footnote ³² At the time of writing, there are 379 COVID-19 vaccines listed as being in development on the WHO COVID-19 vaccine tracker,Footnote ³³ 199 in preclinical development, and 180 in some phase of clinical development or already approved for use. Of the vaccines in clinical development most are injectable (as are the already marketed vaccines), but five are oral and eighteen are intranasal or inhaled, thereby offering potentially significant advantages in terms of mode of administration. According to the UNICEF COVID-19 market dashboard, fifty-eight vaccines have been approved for use in at least one country and eleven have achieved WHO emergency listing, but only a limited number of these are in widespread use.Footnote ³⁴ Of those, 17.8 billion doses have been purchased or ordered in the open market, but more than 15 billion of these come from only four developers: Pfizer/BioNTech, AstraZeneca, Moderna, and Novavax. Pfizer/BioNTech alone accounts for 7 billion doses. These numbers underestimate the importance of vaccines developed and used in China and Russia. They have been widely used in China and Russia, respectively, but have had little success in the market. The first approvals of the AstraZeneca and the Pfizer/BioNTech vaccines for clinical use were issued in December 2020.

This data allows us to consider the preemption and social value issues in more detail in relation to the schematic argument presented in the “Background” section above. At the start of the COVID-19 pandemic an effective COVID-19 vaccine would clearly substitute for the placeholder X in P5 (“X is a medical intervention that can reduce the net negative consequences of the COVID-19 pandemic”) and any individual vaccine under development based on a reasonable vaccine technology would substitute for X in P5* (“X is a medical intervention that can potentially reduce the net negative consequences of the COVID-19 pandemic”). We might even be willing to replace “can potentially” with something like “is likely.” However, reflecting on the data about COVID-19 vaccine development, we should probably rethink this last change. Of the 115 vaccines in development in April 2020, only a few completed development and, of those that completed development and were approved for clinical use, only a few have actually been produced and used at any significant volume (as shown above). Part of the reason for this is preemption. As soon as one effective vaccine is available the need for developing another objectively decreases, as does the incentive for both commercial and government actors to do so. The first approved vaccine (or vaccines, if more than one is approved at roughly the same time) is likely to be most used, even if better vaccines can be developed. The Pfizer/BioNTech vaccine is not perfect, because it does not provide 100 percent protection and the protection wanes over time, but it is good enough to diminish significantly the incentive to develop other vaccines, unless they offer something significantly different, for example, a different mode of administration, better logistics, or longer-lasting or lifelong protection. The research and development leading to the Pfizer/BioNTech vaccine clearly has generated very large social value ex post. This vaccine has saved many lives and prevented much suffering. But ex ante it was in competition for generating this social value with at least 100 other vaccines in development and most of those other vaccines never produced any social value apart from the knowledge generated through their ultimately unsuccessful development.

Partial preemption of social value can also occur between different types of interventions. If vaccines had not been developed, then the social value of effective social interventions and drug treatments would have increased manifold, as has until now been the case in relation to HIV/AIDS where promotion of condom use and safe-sex practices as well as effective anti-retrovirals for treatment and preexposure prophylaxis have delivered almost all of the social value following from HIV/AIDS research.

These considerations show that the link between (1) an intervention X being a possible effective intervention in relation to a particular public health crisis and (2) X and a fortiori research on X creating significant social value is much more complicated than it initially seems. Even in situations where X is proven by research to be effective, there is no guarantee that any social value will eventuate.

This skeptical argument concerning the predictability of a straightforward link between research, innovation, and social value for particular interventions can be criticized. A first possible criticism is that the argument ignores the possibility that a particular intervention can have a very long “tail,” that is, that the intervention will continue to be used for a long time and therefore continue to accumulate net social value. A specific instance of this could be the Pfizer/BioNTech vaccine that will continue to be used for elderly and at-risk populations. The counterargument is that whereas it is true that some interventions will have long tails, this is no more predictable in relation to a particular intervention at the research stage than is whether that particular intervention will be effective and, if effective, will be widely used.

A second possible criticism is that it is problematic to look at specific interventions in isolation; the real connection between interventions, research, and social value is rather at the level of classes of interventions. A specific instance of this would be to move the focus from individual vaccines to classes of vaccine technologies. The mRNA technology that is the basis for both the Pfizer/BioNTech and the Moderna vaccines can, for instance, be used to develop vaccines for other infectious diseases and potentially also for vaccines that protect against various forms of cancer. The counterargument is again that whereas it is true that some new vaccine technologies can ex post be identified as having wider use and thereby significant social value, many novel vaccine technologies were researched and it was ex ante unpredictable which of these would be effective and produce large social value as technologies.

Justifying research ethics exceptionalism: Research ethics is an obstacle to research

The other important part of the schematic argument for changes in research ethics is represented in P6–P6**. The claim here is that research ethics is an obstacle to research, whether by making it more difficult to do research (P6), by slowing down research (P6*), or by making effective research prohibitive (P6**).

P6* is plausibly true. We have no good reason to believe that our research ethics systems are effective in processing applications as quickly as they could. In relation to a single research project it is true, in most cases, that research ethics slows down research, if nothing of a productive nature happens between the time of submission for research ethics approval and the time when approval is given. The project could start earlier, if researchers did not need research ethics approval or the research ethics system was faster. There is thus a crisis-independent argument for streamlining the research ethics approval process and working to minimize the time it takes to process submissions, if we accept that most research projects generate some social value. How much a system should be streamlined would then depend on (1) it being able to maintain the quality of its processes and (2) the balance between the increased costs of running a more streamlined system and the time-saving benefits generated.

This general line of argument for more efficient research ethics processes can, however, only justify exclusive fast-tracks for COVID-19 research, if we assume either (1) that all COVID-19 projects are more important than all other research projects or (2) in “social value” terms that because we are in a public health crisis, it follows that all research projects related to relevant interventions are likely to generate more social value than all other projects and/or are more urgent. These assumptions are, however, obviously false. The expected social value and urgency of COVID-19 research will be distributed across a wide spectrum, as will the social value and urgency of other biomedical research; these spectra will have a considerable overlap and might actually completely coincide. Research into any of the diseases and conditions that cause significant loss of disability-adjusted life years globally will not suddenly become unimportant in a public health crisis.

There are two possible counterarguments available to proponents of COVID-19 fast tracks and, more generally, public health crisis fast tracks. The first counterargument is based on the premise that research ethics systems had to adapt rapidly to the COVID-19 pandemic; it was, after all, a sudden crisis and they were therefore justified in basing some of their adaptations on simple rules of thumb. It might have been better to evaluate the expected social value and urgency of each individual research project, but in a crisis situation a decision made on the binary criterion of whether a research project was related to COVID-19 or not was good enough in relation to fast-tracking.

The second counterargument is based on urgency. The basic idea is that although there is an overlap in social value between COVID-19-related and other research, COVID-19-related research was at the start of the pandemic much more urgent and should therefore be prioritized. That is, it was more important to get COVID-19-related evidence quickly than to get evidence concerning other diseases and conditions.

These counterarguments are not unreasonable, although the second relies on the somewhat problematic implicit premise that there is no other current health problem where research has the same urgency. Both counterarguments do, however, commit us to acknowledging that if fast-tracking COVID-19 projects leads to “slow-tracking” of other projects or if we completely stop processing non-COVID-19 projects for a period of time, we will create a net loss of value from the research that is approved. COVID-19 projects of little value will be processed and approved ahead or instead of other research that is of higher value.

Let us move on to considering P6 and P6**, which both claim that there are research projects that can generate large social value (following from P1–P5) that are made more difficult (P6) or prohibitive (P6**) by substantive research ethics principles or rules. Are these premises true? Yes, they are true. Some research projects would be made easier if we, for instance, were to change our rules on compensation or payment to participants. Some would only become possible if we were to allow conscription to research, in other ways weaken requirements for voluntary participation, make it more difficult to withdraw, or abolish or weaken limits on research-related risk. However, we know from the history of research that we have good reasons for having these principles, as they inter alia prevent research likely to harm the participants and they prevent the exploitation of research subjects in the pursuit of social goods. We therefore need equally good reasons to weaken or abolish these principles. This is not an example of status quo bias; contra Chappell,Footnote ³⁵ these principles and rules are not traditional or historical in the sense that they have no other justification than having been put in place at some time in the past. This is, instead, a requirement to provide sufficient reasons to change well-justified principles.

The following section will focus on the justification based on the large social value generated by research, and will therefore for the sake of argument assume that there are one or more promising interventions we can identify ex ante as being likely to have large social value that is likely to be actualized and where research is a necessary step in the causal chain leading to the actualization of that large social value. The problem, here, is that the protective research ethics principles we are considering changing not only protect the experiential interests of the research participants, for example, their interest in not being harmed. They also protect their right to being treated as equal citizens in society.Footnote ³⁶ We, for instance, insist on voluntary consent because conscription to research would make some persons mere objects to be used by society in the pursuit of the social good. Similar arguments apply to some of the suggested modifications in relation to research risk and risk payments. Chappell is again forthright:

Here, Chappell assumes that because participation in risky research can be against the interests of participants and that we might therefore find it difficult to recruit anyone, this problem can be solved by changing the incentive structure to offer “whatever level of payment is necessary to make participation worthwhile for them.” At the limit, this presumably applies to research that is known to be lethal. There will be some sum of money that will “make participation worthwhile” for at least some persons. However, any society offering that sum of money would explicitly be communicating that some citizens are worth less than others because we can pay them to die for the rest of us. We would not even be able to rely on “The old Lie: Dulce et decorum est Pro patria mori [It is sweet and fitting to die for one’s country].”Footnote ³⁸ Participants who die or are injured in risky research would not die or be injured due to thinking that it would be fitting to sacrifice themselves for their country; they would die or be injured because they had been induced to participate by filthy lucre.

A more charitable interpretation is that Chappell does not want to take his argument to the limit, but that he wants to have some upper limit on research risk that is higher than we currently accept. However, in the article just quoted, this does not seem to be the case. In a footnote, for instance, in relation to the involvement of older participants in human challenge studies, he states:

If Chappell intends this level of risk to be covered by the argument, it would also entail that participation would not be in a project approved by a research ethics committee or institutional review board. After consideration of the risk-benefit ratio of such a project, no committee could approve a project with known high risk but very uncertain benefits.

In other work (with Peter Singer), Chappell argues for “risk parity,” that is, allowing research participants to take on the same level of risk in research as health-care professionals take on in clinical practice.Footnote ⁴⁰ There are, however, important differences between the two situations relevant to the issue of payment for research. Angus Dawson points out some of these:

Another important observation is that health-care workers are not in general paid a risk premium. For example, doctors and nurses working in specialties with high risks are not paid more than those working in risk-free specialties. Psychiatrists and psychiatric nurses who are at significant risk of patient violence are in general not paid more than, say, radiologists and radiographers for whom this risk is much lower. Health-care workers are not in general incentivized to take on specific risks or paid more when the risk level goes up. During the early phase of the COVID-19 pandemic, when work in intensive-care units became much more risky than normal, risk premiums were not offered for each shift taken on. Insofar as it is possible to estimate from wage data, there is no evidence for “hero pay” during the COVID-19 pandemic.Footnote ⁴²

People are paid to take on tasks or jobs that they would not do otherwise. Some of that payment is, in certain cases, a risk premium especially in unionized sectors,Footnote ⁴³ but even in those cases people are not paid for taking on the risk of one particular event. Firefighters may be paid more than we pay other similarly trained emergency workers, but they are not offered extra “event risk” payment for entering a particular risky building on fire. Part of the reason for not paying for event risk is that this is likely to be exploitative and will still be exploitative, even though it makes those who take the payment and survive unharmed better off. Instead of treating people as equals, we decide to exploit already existing inequalities.Footnote ⁴⁴

From analysis of the “research ethics” premises P6–P6** in the schematic argument above, it follows, first, that changes to the research ethics system aimed at making it more effective can often be justified. However, the justification does not follow primarily from the crisis situation. The justification is primarily that there are general reasons for having an effective research ethics system that processes all valuable research as quickly and efficiently as possible. Second, it follows that the mere promise of even very large social value is an insufficient justification for changes to substantive research ethics principles and rules, if the prior justification for these principles and rules is that they protect the basic interests of participants or their basic right to being respected and treated as an equal.

It might be argued that we could professionalize research participation and create a corps of trained research participants who are ready to participate in risky but necessary research when society faces an existential public health crisis. The analogy here could be to the most elite units of the armed forces, for example, U.S. Navy Seals or the U.K. Special Air Service. Members of these elite units of the military are trained to see it as their honorable task to take high risks, to kill or be killed in defense of their country. They are also trained to accept that their particular skills may be unneeded for long periods of time. Members of these elite units may not be paid a significant monetary risk premium, but they are paid in the coin of meaning and therefore not exploited. The problem with this suggestion is that whereas it is reasonably clear what mix of special units and military skills a country might need, given the military posture of its most likely enemies, it is much more difficult to predict what types of research and therefore research participants are needed for the next public health crisis.

The power of definition

The argument that an exceptional crisis justifies exceptional measures was widely used to justify policymaking during the COVID-19 pandemic, in research ethics and more generally. This raises interesting questions about the power to define or decide that a particular problem or issue constitutes an exceptional crisis and about the effects of entering a “state of exception.”

At the international level, the World Health Organization (WHO) has the official power to declare a “public health emergency of international concern” (PHEIC) in accordance with the International Health Regulations (IHR).Footnote ⁴⁵ A PHEIC can only be declared in relation to an infectious disease outbreak and is defined as “an extraordinary event which is determined, as provided in these Regulations: (i) to constitute a public health risk to other States through the international spread of disease and (ii) to potentially require a coordinated international response.”Footnote ⁴⁶ A PHEIC is declared by the Director-General of the WHO on the advice of the IHR Emergency Committee and PHEIC status is reviewed every three months. Declaring a PHEIC has two main functions. First, it signals to the international community that a significant international outbreak of infectious disease is occurring. Second, when a PHEIC has been declared, the WHO Director-General can issue “temporary recommendations.” These are not legally binding, but they can include recommendations of health measures states can implement to reduce or prevent further international spread of the disease and recommendations about how to avoid unnecessary restrictions on international trade and travel. The recommendations are intended to guide state actors in order to facilitate a coordinated and proportionate international response to a given public health crisis.Footnote ⁴⁷ There has for a long time been significant debate about whether the PHEIC system works and actually leads to international coordination; this debate has intensified after COVID-19.Footnote ⁴⁸ Since the current version of these regulations came into force, the WHO has declared such an emergency in relation to: Swine flu (2009), Polio (2014 ongoingFootnote ⁴⁹), Ebola (2014–2016), Zika virus (2016), Kivu Ebola outbreak (2019–2020), COVID-19 (2020–2023), and MonkeypoxFootnote ⁵⁰ (2022–2023).

An analysis of the rationales of the IHR Emergency Committee in relation to cases from 2009–2020 where a PHEIC was declared and cases where a PHEIC was considered but not declared has shown that the IHR’s three criteria are not consistently applied by the Committee.Footnote ⁵¹ The COVID-19 pandemic was, for instance, declared a PHEIC without the Committee deciding formally that it constituted “an extraordinary event” and, in the cases where a PHEIC was not declared, this particular criterion was not explicitly mentioned by the Committee. It thus seems that what does most of the work in relation to the WHO’s IHR Emergency Committee PHEIC decision-making is the risk of international spread requiring a coordinated international response.

Because of the problems and weaknesses identified in the IHR and the PHEIC process, including the fact that the recommendations issued are not legally binding, the WHO member states are in the process of negotiating changes to the IHR. A new legally binding pandemic treaty is also being developed,Footnote ⁵² although there is no certainty that member states will be willing to cede any significant power to the WHO.Footnote ⁵³

Given that the WHO currently has no power to issue legally binding directions to states in relation to public health crises, a more general issue arises about the legal declaration of a state of emergency by actors that have the power to change the legal landscape. The most obvious members of this class of actors are nation-states, but the class also includes supranational bodies such as the European Union and subnational entities such as states in federated nation-states or lower subnational entities with public health responsibilities (for example, municipalities). The legal and philosophical aspects of declarations of “a state of exception” were extensively discussed in the 1930s and 1940s in response to the rise and fall of new forms of political arrangement in Europe.Footnote ⁵⁴ This topic has recently been reinvigorated in the work of Giorgio Agamben.Footnote ⁵⁵ Agamben provides a historical analysis of emergency regimes from early Roman law to the present day, focusing primarily on emergencies caused by internal and external threats to political stability. He takes his point of departure in Carl Schmitt’s famous observation that “[s]overeign is he who decides on the exception.”Footnote ⁵⁶ Agamben argues that the possibility of declaring something an exceptional emergency situation provides the political decision-maker with the opportunity to create a state of exception. The state of exception creates a new “lawless” space that can then be filled by the sovereign decision-maker.Footnote ⁵⁷ The far-reaching political conclusions Agamben draws from this analysis and from his claim that we currently live in a permanent state of exception are beyond the scope of this essay, though, as is his own contentious analysis of COVID-19 public health interventions.Footnote ⁵⁸

However, Agamben’s general point that emergencies open up a new space for radical decision-making by those with the definitional power to declare a state of exception is worth exploring in relation to research ethics exceptionalism. To what extent does the “fact” that something is a public health emergency warrant or justify the suspension of normal research ethics principles and their replacement by a potentially radically different research ethics? It is obvious that an Agambenian sovereign can suspend normal research ethics principles, but would it be justified?

As noted above, our current research ethics was developed as a response to problematic research practices during times of crisis. It is therefore to a significant extent protectionist and categorical. This is evident, for instance, in many of the ten statements of the Nuremberg Code. Statement 1 of the Code is an exceptionless commitment to voluntariness and consent: “The voluntary consent of the human subject is absolutely essential.” Statement 5 also equally categorically states that “[n]o experiment should be conducted where there is an a priori reason to believe that death or disabling injury will occur; except, perhaps, in those experiments where the experimental physicians also serve as subjects.”Footnote ⁵⁹ These requirements are not stated as optional or relative to a particular context of normality. They are explicitly designed to put limits on research in a time of existential, national crisis. They are thus analogous to those articles in human rights conventions that do not have exceptions and do not allow state parties to enter derogations. In the European Convention on Human Rights the emergency provision in Article 15, for instance, allows for derogation from most articles of the Convention “[i]n time of war or other public emergency threatening the life of the nation,” but explicitly prohibits derogation from Article 2 on the right to life; Article 3, which prohibits torture and inhuman or degrading treatment; Article 4.1, which prohibits slavery and servitude; and Article 7, which prohibits retrospective punishment.Footnote ⁶⁰

If we understand the most foundational research ethics principles in this way as nonderogable, it raises the question of which principles belong to this class. A full analysis of this question is outside the scope of this essay, but an initial suggestion might be that statements 1, 5, and 9 of the Nuremberg Code are core principles.Footnote ⁶¹ They protect the most basic rights of research participants against the state and researchers acting on behalf of the state, and they have been carried over into later international normative documents. Many other aspects of research ethics, including perhaps all of the specific features of the research ethics system as currently constituted, would be outside of the core and therefore in principle derogable in a true emergency. The Agambenian sovereign could thus be justified in suspending the need for research ethics committee approval, but not in suspending the right to withdraw from research or the limit on research risk.