1 Introduction

One of the most enduring debates over COVID pandemic measures – and the debate that may have the most enduring relevance for future public health crises – concerns whether the extreme, coercive, unparalleled mitigation and containment measures implemented were justified, or whether such measures are in principle justifiable. These measures took many forms – but restrictions on gatherings and movement, masking requirements, and heavily incentivized or mandated vaccination programmes played a key role in mitigation and containment across many countries. Each of these measures involved an – sometimes quite significant – imposition on liberty; a value not generally thought to be easily, justifiably violated under normal circumstances. Was there something about this particular set of circumstances that might provide justification for such an imposition?

There was one distinctive feature of the COVID pandemic, which may be significant when we are searching for a justification for these extreme measures. Often, when we look at the harms wrought by a public health crisis, such as a measles outbreak, we are focused on the loss of health and life that result from infection. The COVID pandemic, too, threatened many deaths, particularly among the elderly and vulnerable, but there was another sort of potential harm that is not a feature of many other disease outbreaks: the fact that, in many countries, the insufficiently constrained spread of COVID stretched healthcare systems to and beyond capacity. In some ways, this is an obvious point. The threat of overwhelmed health care systems was well-recognised by policy-makers. It formed the primary impetus for the imposition of lockdowns in various countries (UK Government 2020; Government of the Netherlands 2021; The Economist 2023), as well as the chief justification, in Austria, for their controversial vaccine mandateFootnote 1 (Druml and Czech 2022; Schuetze 2022). But the distinctive nature of this harm was not well-recognised in ethical work on the justifiability of coercive public health measures during COVID. This is a problem, because the route that ethicists often took in searching for justification is both unable to capture the full scope of the harm done when healthcare systems are put under such strain, and leaves alternate, potentially stronger and more fitting sources of ethical justification for coercive public health measures on the table. Because the threat of institutional failure might be a feature of a variety of future crises – both in public health and beyond – understanding the nature of this harm, and debating what measures might be justified to curtail it, might help us better prepare for future crises.

I will draw this out in three parts. First, I examine how many ethicists have attempted to assess whether coercive measures like lockdowns and vaccine mandates are justified. I point to a central thread in these arguments – their reliance on John Stuart Mill’s “harm principle” to delineate justifiable bounds for coercive measures. Though this approach has some significant strengths, I suggest that there are also various problems with this approach, chiefly, that this type of argument may lead us to focus on the type of harm that individuals pose to other individuals – a place where the principle functions in the most clear and straightforward manner. I show what this type of approach might miss in the second section, where I draw out three ways in which a focus on individual harms cannot capture the full scope of the harm threatened by overwhelmed healthcare systems. I will then sketch some implications that a focus on broader societal harm might have for the justification of restrictive public health policies.

2 The harm principle – and individual harm

2.1 Background: the harm principle and mandatory vaccination

A public health crisis can cause widespread harm. But we generally don’t think that minimizing harm is our only ethical imperative, even when the threat of harm is significant. How do we determine what we can justifiably do to limit the harm of a public health crisis? One influential answer to this question comes from the ethical debate on mandatory vaccination – a discussion predating the COVID pandemic.Footnote 2 Flanigan (2014) and Brennan (2018) both suggest that we should start with the John Stuart Mill’s “harm principle” – the stipulation that “the only purpose for which power can be rightfully exercised over any member of a civilized community, against his will, is to prevent harm to others.” (Mill 1869; p. 22). Both endorse this starting point for similar reasons. Flanigan claims that the “minimal condition” (2014, 17) “that coercion is justified to prevent people from harming others and imposing serious risks on others” (16–17) can be endorsed by “conservatives, egalitarians, libertarians, and even anarchists” (17). Brennan similarly suggests that even a libertarian – a holder of the view that “that each individual is endowed with an extensive set of strong rights against interference…and, as a result, the permissible scope of government is quite limited” (2018, 37 − 8) – would agree that state interference is warranted when an individual imposes “a serious risk of harm upon other people” (39). By making an argument that even those most sceptical of government interference would accept, Brennan aims to make the strongest possible case for coercive measures. Both Flanigan and Brennan, that is, aim to take as their starting point a justification for coercion that few would reject. And it is certainly true that the harm principle is widely regarded as providing a strong justification for coercive interference, both in public health ethics (Childress et al. 2002) and beyond (Feinberg 1984).

The core of Flanigan’s harm-principle based argument is a simple analogy. Just as you are prohibited from shooting a firearm in the air, she argues, so should you be prohibited from going around unvaccinated. When you shoot a bullet in the air, you may not intend to cause any harm, you may not be aware that you could cause harm – but your conduct poses an unacceptable risk to others. By failing to vaccinate yourself against, let’s say, measles, and then interacting with people in various settings, though you may not foresee or intend becoming infected and infecting someone else, your conduct could be deadly to those you risk infecting. We should regard this as a violation of the harm principle – and behaviour that we thus might justifiably restrict, according to Flanigan, for exactly the same reasons (2014).

One challenge for Flanigan’s argument is to establish that the level of risk posed by shooting a bullet in the air is akin to the level of risk posed by going around unvaccinated. After all, we pose risks to people all the time with our conduct, and if we were to regard any imposition of risk as sufficient to trigger the harm principle, our conduct would be very severely curtailed. Certainly, this would not be in keeping with Flanigan and Brennan’s aim of grounding a justification for coercion on a minimal and widely accepted principle. It is for this reason that explications of the harm principle limit the type of risk imposition that might justify coercive intervention to only “unreasonable risk” (Feinberg 1984; p. 11) or “serious risk” (Childress et al. 2002, 175). The magnitude of the risk determines the strength of a harm-principle-based defence of coercive measures, and as the severity recedes, the force of this argument weakens (Giubilini 2020).

This problem is recognised by Brennan, who suggests that Flanigan’s analogy may be suited to justifying compulsory quarantine measures, but strains credibility when applied to people that are not already infected with a dangerous disease. As he argues:

Quarantines restrict people’s liberty, but they are justified (in certain cases) because these people present a clear and present danger to others, or at least are at a very high risk, as individuals, of presenting a clear and present danger to others. But mandatory vaccination forces individuals who are not a clear or present danger to others nor are at a high risk of being a clear and present danger to accept a vaccine against their will. When a large group of people refuses vaccines, the group may impose a risk, but we cannot easily attribute the risk to any individual within the group. While quarantines focus specifically on dangerous individuals (because they are infected or are likely to be infected), mandatory vaccines are targeted at everyone. But the problem is that individuals as individuals make little difference…vaccination presents a collective action problem, in which individuals as individuals are unimportant. (2018, 3)

Brennan thus attempts to establish that the harm principle also entails an obligation to avoid contributing to collective risk of harm (and that coercion is justified when these obligations are not met). This rests on two arguments. The first is the “clean hands principle”; the claim that that there is a “(sometimes enforceable) moral obligation not to participate in collectively harmful activities” (2017, 4), including the imposition of certain types of risk. Second, in order to ensure that his argument does not justify too much interference by including all risk, he utilises Sven Ove Hansson’s distinction between acceptable and unacceptable risk: “it is acceptable to expose someone to risk ‘if and only if this exposure is part of a social system of equitable risk-taking that works to her advantage’” (2017, 4). Hansson’s principle, as Brennan explains, allows us to explain why we are allowed to expose others to risk through ordinary driving, while we are not permitted to leave a bomb on the street that has a small chance of exploding – allowing the former activity works to everyone’s benefit, while allowing the latter does not. Because refusing vaccination violates the clean hands principle by contributing to the collective exposure of others to unacceptable risk, Brennan claims that coercion is justified.

However, if we are basing this justification on the harm principle, there are problems with both steps of this argument. As Alberto Giubilini points out, though it may be wrong to contribute to a collective harm, “it is not because the contribution is in itself harmful” (2020, 449). Brennan would acknowledge this – it is precisely because individual contributions are negligible in the case of vaccination that Brennan turns to the consideration of collective risk. But in appealing to the clean hands principle, Brennan is appealing to a principle that requires a separate justification. As Giubilini argues; “[i]f enforceability is only justified by harm prevention, then the enforcement of the principle is not justified” (2020, 449). The harm principle alone cannot provide the basis for this argument.

Brennan’s appeal to Hansson’s account of unacceptable risk also poses problems for a harm-principle-based justification – as Justin Bernstein argues, “the distinction between acceptable and unacceptable risk turns on whether there is a ‘net benefit to each individual’ if we allow individuals to expose others to the risk in question” (2017, 794). However, as Bernstein points out, this amounts to the suggestion that “the state is justified in using coercive measures to secure public goods because doing so increases the welfare of each” (2017, 794). This, as Bernstein claims, is at odds with the commitments of libertarianism, which entail that individuals hold strong negative rights prohibiting interference on the basis of promoting others’ welfare. This more controversial claim cannot be derived from the minimal and widely appealing version of the harm principle that Brennan and Flanigan take as their starting point and ultimate source of justification.

The appeal of a harm-principle-based justification for coercive public health measures is clear – if we can make a case for unpalatable, extreme measures on the basis of a principle that even those most sceptical of government coercion would accept, that that “no responsible liberal theorist” would deny (Feinberg 1984), then this justification must be viewed as very robust. However, as we have seen, we run into problems when attempting to utilise the harm principle as the basis for justifying coercion in cases when the individual contribution to the imposition of risk is insignificant. And attempts to expand the principle to cover collectively imposed risk requires supplementation with other, unrelated principles that do not enjoy the wide support of the harm principle. Let us now turn to how the harm principle has been utilised in the debates on COVID measures such as lockdowns.

2.2 The harm principle and COVID restrictions

The harm principle also featured in debates about the justification of COVID lockdowns, but the focus, here, remained squarely on direct and indirect risks that individuals pose to other individuals. In the earliest days of lockdown, there was quite some debate about how we should understand going about our daily business during a pandemic. Alec Wallen and Bashar Haydar suggested that staying home during the pandemic amounted to “saving lives”, and though we may have some moral obligations to positively assist others in this way, these should be understood as limited (2020). Helen Frowe responded by contending – much along the lines of Flanigan – that this misconstrues what we are doing when we limit our contact with others during a pandemic. By staying at home and refraining from infecting you with COVID, Frowe argues, “I refrain from harming you” (2020), and by going about my usual business under these conditions, I pose you some risk of harm, necessitating a discussion about whether restrictions on this conduct might be justified. Fiona Woollard largely agrees with Frowe on this point, contending that going out and infecting someone should be regarded as harming them “on any account of harm”. For Woollard, this means that “governments may be permitted or even required” to prevent this imposition on others (2020).

Further into the pandemic, when COVID vaccines were developed, Stephen John contemplated whether Flanigan’s harm-principle-based arguments could also be marshalled in support of mandatory COVID vaccination. Even if these arguments succeed in the context in which they were originally employed, he argues, they do not carry over to COVID in a straightforward way, because COVID vaccines do not have such a profound preventative effect transmission as, for example, measles and pertussis vaccines (2021). Even if we were to regard going about unvaccinated as posing sufficient risk of harm to others to trigger the harm principle, it is not the case that being vaccinated against COVID provides an effective means of preventing that harm. Perhaps, John then considers, we can regard failing to be vaccinated against COVID as posing an indirect risk of harm to others. That is, by failing to be vaccinated, you put yourself at higher risk of infection, severe illness, and hospitalization. Once you are hospitalized, you deprive others of limited medical resources, who may suffer serious harm or death as a result (2021). Frowe also suggests that risking infection and hospitalization by failing to limit contact with others poses this sort of indirect harm to other individuals, and should at least lead us to consider whether this amounts to the unacceptable imposition of risk. Woollard also notes another type of indirect harm posed by going about one’s business during a pandemic - by infecting someone else, who then infects their family, we create chains of infections, hospitalizations, deaths, that “grow exponentially” (2020).

John ultimately rejects this line of reasoning too – to expand the harm principle in this way, he contends, would justify all sorts of incursions into the lives of others. Failing to take adequate care of one’s own health by eating fatty foods, for example, usually regarded as a primarily self-regarding decision, could also leave one open to the accusation that one risks harming others by increasing the probability of requiring scarce medical resources (2021). If we see the appeal of harm-principle-based justifications, as Brennan and Flanigan do, in providing a minimal and widely accepted route for justifiable incursion into the liberty of others, we should also be concerned to see the principle employed in this expansionist and controversial way. Ultimately, John concludes, running up against yet another familiar problem, it is doubtful that the magnitude of the risk posed to others through failure to be vaccinated against COVID can be regarded as sufficient to trigger the harm principle.

The upshot of all this, for our purposes, is that the debate on justifiable coercive COVID measures and the harm principle remained in the domain of the risk of harm that individuals pose to other individuals. It may be, particularly when we consider lockdown restrictions, that appealing to the direct risk of harm that individuals pose to other individuals is where the principle is on the most solid ground – if a virus is particularly serious and widespread, the risk of harm posed to others by failing to stay home might be of sufficient magnitude to make a compelling case for the deployment of the harm principle, and attendant coercive restrictions. We may also need to price in the indirect risk we pose to others through chains of transmission. A harm-principle-based argument for coercion to prevent people from using limited medical resources is, for the reasons canvassed above, on shakier ground.

But even if it were to succeed, none of these individual-focused arguments capture a distinctive threat posed by the COVID pandemic, and the primary impetus for the imposition of restrictive measures like lockdowns and vaccine mandates: the threat of overwhelmed healthcare systems, and the harms that ensue, including the disruption or suspension of prevention and treatment services (World Economic Forum 2023), and the failure to provide patients with treatment due to resource shortages (Khetpal 2021). In order to fully capture these harms through the lens of the harm principle, we would need to extend it to collectively imposed risks, but, as we have seen above, the harm principle alone does not provide good justification for coercive measures in cases of collectively imposed risk. It is perhaps for this reason, coupled with the wide appeal of the harm principle, that the philosophical discussion on the harms of COVID and the justification of coercive measures have focused so heavily on the individual harms that the principle most easily captures.

But these broader harms should not be neglected – if we do not understand the nature of these harms, we do not understand the threat posed by this sort of public health crisis. And once we understand the nature of these harms, it may open up more compelling and fitting sources of justification for coercive measures. And although the harms resulting from overwhelmed healthcare systems are not a feature of the public health crises that receive the most attention in public health ethics, such as measles outbreaks, nor are they unique to COVID – the diversion of resources to tackle the Ebola virus crisis in Liberia between 2014 and 16 led to thousands of deaths from preventable diseases as patients were unable to access care (Venkatesan 2020). This type of threat may also feature in future pandemics. In what follows, I will catalogue the harms posed by severe strain on healthcare systems, before sketching the implications of recognising these harms.

3 Not just a matter of individual harm

3.1 The “doom loop”

The first way in focusing on the harm that individuals pose to other individuals fails to capture the full extent of the harm resulting from overburdened health care systems is the ability to generate what we might refer to as a “doom loop” (Edwards 2023). During some periods of the COVID pandemic, we saw a significant diversion of resources to COVID patients – for example, through the dedication of a certain proportion of staff or hospital beds in anticipation of an influx of seriously ill COVID patients. This diversion of or lack of resources led to the postponement, for example, of cancer screenings. By the time that patients are screened, identified, and treated, their condition is likely to be more serious (Cancino et al. 2020). This will require more and lengthier care, expending more resources than would have originally been needed, and causing lengthier waits for subsequent patients. This “doom loop” leads to worse outcomes with every iteration and, once the situation spirals past a certain point, it can be very difficult, perhaps even infeasible, to reverse the trajectory (Edwards 2023). The UK’s National Health Service may present an example of this, with a backlog of patients waiting for care that has climbed steadily since the beginning of the pandemic, and shows no signs of abating (BMA 2023). These compounding, runaway effects are not adequately captured by viewing this, as we have seen above, as one individual “using up” resources that could go to another.

3.2 Ramifications of institutional failure

A focus on whether one individual is harming another individual also fails to capture the wider societal ramifications of the failure of a central social institution. A failure to provide adequate healthcare has a ripple effect across many domains of society. It can generate negative economic effects – increased absence from work, earlier retirement – and a resultant decline in economic productivity. Demand on scarce healthcare resources increases the likelihood – or exacerbates the effects – of a “two-tier” healthcare system, in which better care is purchased by the rich, while the worst-off are left to rely on the poorly-functioning publicly accessible system (World Economic Forum 2023). These same groups are also likely to face a disproportionate disease burden (McGowan and Bambra 2022). The failure of one institution can lead to other institutional failures, exacerbate inequalities, and have other broader negative impacts that it can be difficult to fully appreciate when we are attempting to pinpoint whether individuals are causing other individuals harm through their personal conduct.

3.3 Psychic costs

The third reason is more difficult to quantify, but important nonetheless. In societies with a robust public health care system, the existence and universal accessibility of this system plays an important social role. Nowhere is this more evident than in Britain (Higgins 2023), where the existence of a national health service is touted as an exemplification of central societal values – providing high quality healthcare to all, on the basis of need (NHS England 2023). When a healthcare system ceases to function properly, you can no longer rely on the idea that you will receive help when you need it (Edwards 2023). Beyond the obvious direct harms that result when patients cannot access needed medical care, this can have concrete, wider impacts – for example, decreased confidence in public health institutions might contribute to the proliferation of misinformation, and a lack of trust in public health measures like vaccines (World Economic Forum 2023).

But there is more to it still – when a central social system, upon which one is used to relying, ceases to function, one can come to have the feeling that society is falling apart (The Economist 2023). It breaches the covenant between the individual and society – that, in exchange for your contributions to society, when you are in distress, regardless of your position, your will receive help. And it undermines a sense of solidarity in the provision of health services – rather than a system which is contributed to, and provides for, all, severe resource shortages force the interests of different groups into stark opposition, as, for example, when cancer or heart disease patients have their care suspended in order to provide urgent intensive care.Footnote 3

Furthermore, and somewhat less tangibly, seeing and allowing the suffering of others – patients turned away from hospitals, unable to access medical care, suffering severe adverse effects while waiting for emergency help – can have “psychic costs” (see Casal 2007; Dworkin 1988; Feinberg 1986; Le Grand and New 2015). Dworkin and Feinberg illustrate this idea through an argument in support of mandatory motorcycle helmet laws. Though a liberal approach to this issue might suggest we should leave wearing a helmet up to the individual, and – in order to ensure that this self-regarding decision does not impose significant costs on others – refusing to come to their aid if they are involved in a horrific crash, this imposes a significant cost on us, putting us in a position of “ignoring and abandoning those in distress” (Dworkin 1988; p. 127). Feinberg suggests that it would be “unthinkable that we leave the reckless, bareheaded young motorcyclist to die in a pool of his own blood” (1986, 140). We see this line of thinking echoed in discussions on COVID measures – the idea that we should allow individuals to “opt out” of vaccination requirements, and “pay the price” by refusing to provide them care if they later need it, is regarded as “morally monstrous” (John 2021; see also White 2021; Bradley and Navin 2022) – an option that cannot be entertained.

The situation that we are considering here differs slightly – we are not considering cases in which people are abandoned and left to suffer or die because of a decision they have made, but because of an overall lack of resources. But this still amounts to a situation in which we are forced to ignore or abandon those in distress, and thus a situation which could impose severe costs on individuals left to see and tolerate suffering. This may also have a broader effect on “society’s moral environment” (Casal 2007; p. 323). This links to the previous point – a society that does not come to the aid of those who need it, that is not able to uphold the values that it purports to espouse at the most fundamental level, has an effect that is difficult to quantify, but that should not be overlooked.

4 Implications

A focus on the harm principle steers us towards focusing on harms that the principle captures most easily, and where it is able to provide the strongest justification for coercive measures – cases in which individuals pose a serious risk of harm to other individuals. But arguments which focus on the risk of individuals harming individuals – either directly, through transmitting a dangerous disease, or indirectly, by taking up healthcare resources that have been used by others – do not adequately capture the societal harm done when a central social institution ceases to function. And examining the broader societal implications of and difficulty in reversing institutional collapse provides us with alternate paths for approaching this type of threat, and what might be justified to curtail it.

One possible path forward would be to look for ways to supplement the harm principle with other sources of justification for coercion (see Giubilini 2020). But another route might be to take a closer look at arguments that focus on the broader, societal harms posed by this type of crisis. For example, we may begin to regard such institutional collapse as a type of existential threat to the fabric of society. And if this is the case, this might come closer to the kinds of existential threats that Michael Walzer, in the context of the ethics of war, can justify extreme and rapid action (1988, see also Birch 2021; White et al. 2022). Or when we take the broader ramifications of institutional failure into account, a situation in which healthcare systems threaten to be overwhelmed might rise to a level of “catastrophe” that allows, according to many philosophical understandings of the precautionary principle, for us to take drastic action, even when the precise nature of the threat we are facing is uncertain (see Gardiner 2006; Steel 2015). The ability to justifiably act under uncertainty will likely be a necessary condition for drastic action when faced with a rapidly developing public health crisis.

This is not to say that either of these potential routes of justification are at all well-developed, or ready for deployment in this context as a basis for justifiable public policy. A precautionary approach guarding against catastrophe, or a situation approaching, as Walzer would call it, “supreme emergency” should not provide us with carte blanche. Either of these routes for justification must be deployed with extreme caution, and we must consider, carefully and at length, exactly what may be justified under conditions of broader institutional failure, and its wider ramifications on society without trampling individual rights and liberties. But this is precisely why now – before the next rapidly moving crisis hits – is the perfect time to begin to explore these approaches, which have in mind the type of broad societal harm which the failure of healthcare systems threaten. These approaches are not adequately developed to provide us with nuanced guidance, a path that balances upholding civil liberties with effective response to a threat.Footnote 4 But they focus more squarely on the types of harm that we face when an institution threatens collapse, providing us with a starting point that does not tempt us to mischaracterise or miss a dimension of what we face. The COVID pandemic has revealed the paucity of philosophical resources apt to provide guidance in this type of situation, and the harm principle, despite its attractions, is not well suited to either capture or provide guidance on a situation that involves this type of broad, societal harm. Starting with, and examining the nature of the harms wrought by the failure of a central institution opens up other avenues of inquiry for methods of justification and guidance for intervention in a crisis.

Another rather different implication that this argument may have concerns the ongoing, slow moving crisis facing health care systems such as the NHS. As noted above, the universal health care system in Britain is under continued severe strain. Although some progress has been made on reducing lengthy wait times for ambulances (BMA 2023), which ballooned to more than 90 min for patients suffering from strokes or heart attacks earlier this year (Edwards 2023; The Economist 2023), patients who have not received care for various ailments are continuing to flood the system, sicker and in need of more care, due to delays. More than a tenth of the UK’s population is waiting for non-emergency medical treatment (World Economic Forum 2023), and despite efforts to tackle the backlog (England 2022) the number continues to climb (BMA 2023).

This is not the sort of sudden shock presented by a global pandemic – the reasons behind the slow failure of the UK’s health care system are numerous, chronic and complex. The strain of the pandemic certainly played a significant role, but chronic underfunding (Higgins 2023) meant that the pre-pandemic system was already faced with a significant backlog of patients (BMA 2023) and was less robust against the pandemic shock. Aging populations and inflation are also contributing to the dire situation (Edwards 2023), and underfunding of related social systems, like social care, are leaving many patients unable to be discharged because they have nowhere to go (Helm et al. 2023). The measures to address this sort of crisis are very different to what we might entertain against a sudden pandemic – adequately addressing these factors will require difficult, systematic and long-term reform. But the effects of such a chronic crisis are the same as what has been outlined above. If these effects justify drastic measures like lockdowns, they justify approaching the difficult task of restoration and reform with the same urgency.

5 Conclusion

The harm principle has many attractive features – almost no-one would deny its validity, and it provides us – in theory – with a clear dividing line between justifiable coercive intervention, and a sphere in which coercion cannot be justified. But determining precisely what counts as harm towards others can be fuzzy in practice. As harms become more indirect, or as it becomes more difficult to identify each individual’s contribution to an even very severe collective harm, it becomes more difficult to implement the harm principle, particularly while preserving the feature which cause so many to turn to it in the first place – its wide acceptance, based on the fact that it is a minimal principle, giving maximum scope to individual liberties. This may lead us, as it seems to have in the debate on COVID harms and the justifiable measures to curtail them, to miss a broader and fuller picture of the harms facing us, in favour of zoning in on the harms individuals pose to individuals, where the principle stands on its most solid and compelling ground. The most pressing harms of the COVID pandemic – the ones that formed the basis for public policy action across the world – were broader in nature, and unable to be reduced to individual harm. Taking as our starting point the nature of the harms posed in a public health crisis, when a central social institution is threatened, might provide us with a new and more fitting avenues of justification for coercive measures in these crises. Alternate ethical approaches, which focus on harms like that posed by institutional failure, are sorely in need of further development to provide adequate guidance in these situations, but may better allow us to see understand the broader nature of the threat, and its ramifications.