What the harm principle says about vaccination and healthcare rationing

Abstract

Clinical ethicists hold near consensus on the view that healthcare should be provided regardless of patients’ past behaviors. In classic cases, the consensus can be explained by two key rationales—a lack of acute scarcity and the intractability of the facts around those behaviors, which make discrimination on past behavior gratuitous and infeasible to do fairly. Healthcare providers have a duty to help those who can be helped. In contrast, the COVID-19 pandemic suggests the possible recurrence of a very different situation, where a foreseeable acute shortage of healthcare resources means that some cannot be helped. And that shortage is exacerbated by the discrete decision of some to decline a free, safe, and highly effective vaccine, where the facts are clear. In such a future case, if healthcare must be denied to some patients, rationers who ignore vaccination status will become complicit in externalizing the consequences of refusing vaccination onto those who did not refuse. I argue that giving the unvaccinated person healthcare resources that would have otherwise gone to other patients is to wrongfully set back the interests of, or harm, those patients. The article considers rejoinders around the voluntariness of the vaccination choice, which impinges both access and information, and how to scale this criterion proportionally with other rationing criteria that serve utility. Ultimately, the article speculates on why there will be some cognitive dissonance under this approach, while upholding a more general solidarity with and concern for all those seeking healthcare.

I. INTRODUCTION

Those working in clinical bioethics hold a near consensus against what might be called, ‘retrospective moral discrimination in healthcare’ (‘RMDH’).¹ On this view, physicians and other healthcare providers should provide care consistently, without inquiring into the morality or prudence of their patients’ lifestyles or behaviors that precipitated the need for healthcare. Thus, someone brought to the emergency room with injuries from an automobile accident deserves the same care regardless of whether they are the innocent, sober victim in one vehicle or the guilty, drunken driver of the other vehicle.² Skiers with broken bones deserve the same care, regardless of whether they were carefully skiing on the proper groomed slopes or trespassing for thrills off-piste.³ Patients are owed the same cardiac care, regardless of whether one is morbidly obese due to a lifetime of overeating or is a fit and trim marathoner suffering from an overuse syndrome.⁴ In extremis, even terrorists or adversaries in war are thought to deserve the same impartial care as their victims.⁵

In accordance with this near consensus, the field of public health embraces an inclusive ‘ethic of solidarity’, largely rejecting the view that individuals should be held ‘personally responsible’ for their own health, as ‘blaming and shaming’ is neither fair nor effective for maintaining social bonds that are essential for health.⁶ From a somewhat different tack, political liberals also recognize that there is no single metric by which a good life can be evaluated and are thus hesitant to finger-waggingly blame people who choose one way or another.⁷

During the COVID-19 pandemic, US hospitals announced that they must cancel surgeries and other procedures due to an overwhelming number of patients seeking treatment for COVID-19.⁸ And, during the surges when healthcare scarcity was real, those seeking healthcare for COVID-19 were almost all unvaccinated, even after the vaccines were proven safe and effective and distributed broadly for free.⁹ If such a case arises again and difficult choices must be made about who does and does not consume acutely scarce healthcare, one must ask whether a patient’s vaccination status in this situation is a special case, which should be distinguished and excluded from the general consensus against RMDH. The popular press has featured opinions going both ways on this question.¹⁰ In the scholarly literature, one paper relies heavily on the classic RMDH view to argue against vaccine prioritization.¹¹ Another important albeit short piece has provided a framework for considering vaccination as a form of reciprocity.¹² Other peer-reviewed scholarly work has largely ignored or just mentioned in passing this potential criterion for rationing in a pandemic.¹³

Part II of this essay explores two key rationales—lack of acute scarcity and intractability of factual discernment—that support and explain the consensus in those classic RMDH cases. Without challenging the truth of RMDH in classic cases, I ask whether those same concerns apply to vaccination in a pandemic. When there is an acute shortage of healthcare resources that requires rationing at the point of care, and where vaccination status is readily verifiable and interpretable, there would seem to be valid distinctions from RMDH. The field is then open to evaluate vaccination on its own merits.

Part III presents the heart of the affirmative, seemingly novel, argument. I will explore how a person’s decision to not vaccinate in a pandemic, when severe healthcare shortages are foreseeable, creates a risk of harm to other persons. When that harm materializes, healthcare providers must avoid becoming complicit in harm-doing. They should not assist unvaccinated individuals in externalizing consequences onto other persons by displacing their access to healthcare. Accordingly, these considerations suggest that for adult competent adults, vaccination status (or the application of a valid exemption) should be a valid criterion for the allocation of scarce healthcare resources in the pandemic.

This part of the article draws moral insights from analogous areas of law, especially torts. This framework is useful because it focuses on questions of harm, foreseeability, and causation, without aiming to punish (in its primary application, deployed here). I put aside more technical questions of whether and how various healthcare laws and regulations might require or circumscribe the use of vaccination status in rationing.¹⁴

Part IV clarifies the application of the harm principle, in consideration of both fair competition and the act-omission distinction. Part V considers rejoinders around voluntariness of the vaccination choice, which impinges both access and information. We see that the problems are contingent and, in any case, do not override the harm principle.

Part VI briefly considers how to scale this consideration proportionally with other rationing criteria, including those that serve utility. Finally, Part VII considers whether and why there will be some residual resistance to this approach in a deeply seated commitment to social solidarity, contrary to the reasoned analysis presented here. In this part, I explain why the pandemic context is so peculiar and how the slippery slope to outright patient-blaming can be avoided. Part VIII concludes.

Before launching into that argument, let me clarify the scope. First, the impermissibility of RMDH is exclusively because it is focused on past behavior. Nonetheless, the ethical consensus allows that for organ transplants, past behavior (eg an untreated ongoing alcohol addiction) could be ethically considered if it validly predicts future behavior, specifically behaviors that would cause the organ to be wasted.¹⁵ Similarly, there are also forward-looking reasons why vaccination status might be relevant to healthcare rationing. For example, if an unvaccinated person is unlikely to benefit from an intervention, that would be a reason for instead providing the treatment to someone vaccinated.¹⁶ Or, being unvaccinated may actually predict greater marginal benefit, if a vaccinated person is likely to recover without the treatment. Outside the clinical moment, one could make a consequentialist argument that a policy of denying healthcare to unvaccinated people will cause more to get vaccinated ex ante. I am setting all these aside, in part because they rely on contestable and contingent empirical claims.¹⁷ Instead, let us focus on whether the backwards-looking fact that someone presents to a hospital having previously not been vaccinated could be relevant to a rationing decision.

Second, the thesis protects access to healthcare by all persons (vaccinated and unvaccinated) for both pandemic-related and other healthcare, which they would have received in a counterfactual world where there is a pandemic but where unvaccinated persons did not make disproportionate claims on scarce healthcare resources. This thesis says to unvaccinated people that healthcare rationers will try to minimize the externalization of harmful consequences of your decision to decline vaccination, by limiting you to the same levels of healthcare resources in the pandemic that you would have consumed, if you had not refused vaccination. At least in this domain where healthcare rationers must decide who gets what, you may gamble with your life, but not the life of others. Careful attention to this counterfactual emphasizes that the goal here is not to punish or express outrage against those choosing to decline vaccination, but only to cabin their harms.

Third, I repeatedly refer to ‘healthcare rationers’ as whomever performs the gatekeeping role of deciding who does and does not receive healthcare. In some cases, these could be individual physicians acting with professional discretion. But there could also be hospital policies, payer policies, state crisis standard of care policies, or even federal regulations purporting to apply relevant laws (eg the Emergency Medical Treatment and Active Labor Act). The present argument is fully normative, suggesting that whomever is doing this rationing has an obligation to consider vaccination status to avoid harming other patients.

II. TRUE SCARCITY AND FACTUAL INTRACTABILITY

First, social resources are always scarce in a general sense, and there has long been discussion about health policy and the need to set priorities, including the adjustment of insurance rates or the setting of taxes on this behavior or another.¹⁸ Nonetheless, outside a pandemic, the sort of acute or absolute scarcity that requires rationing at the point of care is quite rare in affluent countries like the USA, primarily just applying to organ transplantation.¹⁹ In the other classic and ubiquitous opportunities for RMDH—such as the drunken driver, the off-piste skier, or the overeater—a healthcare provider is in fact able to provide reasonable healthcare to all those who need it. Rationing is not the issue, but rather just moral judgment as to their behaviors.

Several normative lenses show why RMDH would be wrong when no acute shortage requires rationing at the point of care. In terms of professionalism, the healthcare provider’s role is not to judge and smite those who behave badly, but rather simply heal those who need help.²⁰ There is also the idea that we should have solidarity with all persons or that all patients have a ‘right’ to healthcare (even if that right is otherwise unrealized in the USA). Even without fleshing out the basis for solidarity or the source of such a rights-claim, we can suppose that it is not waived by mere imprudent, immoral, or illegal behavior. Perhaps most compellingly, from another perspective, in these routine cases, discrimination against some patients would be gratuitous—it harms those patients, without providing material benefits to anyone else. A simple consequentialist analysis (such as utilitarianism) counsels against RMDH in many of the standard cases.

In contrast for vaccination, at various times during a pandemic, hospitals may face acute (absolute) scarcity, whether of ICU beds, ventilators, drugs, or personnel, and they are accordingly cancelling procedures. In such situations of bed rationing, we may well feel solidarity with every single person who needs healthcare, but that does not undermine the practical necessity of denying a needed treatment to some. Under these conditions of acute scarcity, some patients will be deprived of needed treatments and perhaps die as a result, while other patients will receive the treatment and perhaps survive as a result. The question is not whether to deny patients, but unavoidably, which patients to deny. Thus, the first explanation for RMDH, which explains the vast majority of classic cases, does not fit so clearly with pandemic vaccination with acute scarcity.

Second, when there will be cases of real scarcity, then the question is whether a healthcare rationer can resolve the underlying facts for any rationing criterion that would focus on past behavior. This is a problem I call ‘factual tractability’.

Philosophers debate an approach called ‘luck egalitarianism’, which would hold individuals responsible for the consequences of their culpable choices, unlike those that arise from brute luck, such as the genetic lottery.²¹ But for various reasons, luck egalitarianism is not the mainstream approach in clinical bioethics. As I have argued, the primary problem is that for most practical bioethics and health policy questions, luck egalitarianism is just impossible to apply reasonably.²²

In classic RMDH cases where rationing is actually necessary (such as the liver transplant), the underlying facts are scalar (how much) rather binary (yes or no), and they are difficult to ascertain (exactly when did the liver cancer patient start drinking and in what amounts over the course of her lifetime, and what sorts of evidence can answer these questions). Generally, these determinations must be made in relative terms, comparing the patient’s (alleged) behavior to some normalized benchmark, which then requires ascertainment of whether the behavior was more or less than the customary amount (customary in given times and places where the patient may have lived). Moreover, in cases like alcohol use, there are epistemic questions about what was known or knowable about the risks at the various points in time the patient was making behavioral choices. Linking these difficulties are complex causal questions of fact, since it is not as if alcohol consumption is designed specifically to cause (or prevent) the disease in question, which is instead driven by a host of other behavioral, genetic, and environmental factors not all of which are fully known, generally or as applied to any one case.

If taken seriously by healthcare rationers, as befits a life-or-death decision, these sorts of investigations and determinations would require an entirely different set of personnel, skills, training, and institutional infrastructure, if they are possible at all. These questions are all factual and thus require reliable evidence, including eyewitnesses, expert witnesses, and documents—as well as cross-examination and judgments as to their reliability and other indicia of admissibility. Parsing these problems would make the hospital look something like a court system, with investigators and prosecutors, a body of jurisprudence, and defense counsel (one hopes), along with the weeks, months, or even years to investigate and try such a case. All of that investment would distract from the core healthcare mission and would be utterly infeasible in emergent or even somewhat timely healthcare situations.²³

Thus, lacking the herculean investment and time to perform this moral judgment task appropriately, healthcare providers who nonetheless purported to conduct moral and prudential evaluations on the fly would largely be relying on their ignorant assumptions and subjective preferences. Such highly ambiguous and discretionary decisions are breeding grounds for bias. As Dan Wikler has said, ‘proposals to attach importance in health policy to imprudent health-related behaviour involve a great deal of hand-waving’.²⁴ In the classic cases of RMDH, it is better to not even go down that road.

In contrast, in a pandemic, a competent, adult patient’s decision to get fully vaccinated (or not) is a discrete medical decision, a fact that is either present in the medical records for many patients or readily confirmable in jurisdictions that have vaccination registries.²⁵ States without such registries present a more difficult question, because vaccine cards can be lost (creating false negatives) or forged (creating false positives). Nonetheless, the lack of vaccine registries has not stopped many from advocating (even outside the pandemic context) for vaccine mandates tied to schools, employment, or travel and the implementation of some of these. Further work should explore whether reliance on cards could be supplemented by sworn statements, under penalty of perjury, perhaps backed up by audits. (Moreover, if the remainder of this argument goes through, it will provide a strong basis for creating universal registries so that this morally-salient fact can be more reliably tracked.)

Unlike a patient’s more-or-less, now-or-then practice of drinking alcohol, vaccine status is within the competence and capacity of healthcare providers to determine. The causal connection between being unvaccinated and getting the disease is quite clear and quantifiable, since of course vaccines are designed and tested with randomized trials, specifically for the purpose of quantifying their effects on preventing the specific disease. And we see those effects right in the very same hospital, in the disparate numbers of the vaccinated and unvaccinated presenting themselves for treatment.

For these reasons, the facts around a decision not to vaccinate can be ascertained. Of course, such a determination could be subject to certain well-specified and familiar exemptions, including for patients for whom vaccination is medically contraindicated. One such exemption could be for those who have sufficient immunity due to prior infection.²⁶

So far, I have only suggested that the facts around vaccination can be fairly ascertained by healthcare rationers, unlike the facts around other behaviors on which they may seek to discriminate. Nonetheless, there will be questions about how to morally evaluate the decision to not vaccinate. Some will yield relatively straightforward exemptions, including for those such as children, who were not competent to make an intelligent choice about whether to accept the vaccine. These sorts of factors are also squarely within the competence of healthcare providers to evaluate, since they make similar determinations routinely in evaluating competence to accept healthcare.

More broadly, some may seek to question the voluntariness of the vaccination decision, arguing that in a particular time and place, there is not enough access to the shots or that there is too much misinformation to hold people accountable for their vaccination decisions. These issues are best addressed below (Part V) after laying out a moral framework for vaccination, in the next part.

III. COMPLICITY IN FORESEEABLE HARMS

Together, these two points—lack of acute scarcity and factual intractability—explain the ethical consensus against RMDH in the classic situations, where deprioritizing healthcare for ‘bad’ patients would be gratuitous and infeasible to do reasonably well anyway. But, as we have seen, these considerations do not apply nearly so clearly to vaccination in a pandemic, where scarcity is real and facts are clear. In my view, without undermining its general status, we have largely neutralized the force of the consensus against RMDH as applied to the case of pandemic vaccination.

But we should go further than neutrality if there is an affirmative, moral argument in favor of deprioritizing healthcare for unvaccinated people seeking care for the very same illnesses that the vaccine would have prevented or at least mitigated the need for so much healthcare, if deprioritization can protect the access to healthcare for other persons who would have received that access, but for others refusing to vaccinate. This argument arises from the moral nature of the vaccination decision, as one that can be fairly viewed as consequential. This argument has three subpoints, recognizing that being unvaccinated (i) risks harm to others, (iii) which are foreseeable, and (iii) which are perpetrated through the healthcare rationing.

This argument demands clarity about whether and how we should judge the decision to not vaccinate under the specific circumstances of a pandemic where it is proven safe and effective and offered conveniently for free. As noted, many ethicists embrace political liberalism, recognizing that there is not a single overarching conception of the good life, and that individuals have a liberty to choose whatever path they may prefer, within certain bounds.²⁷ In the classic cases, this liberalism brings a normative buttress to the consensus view against RMDH, making it not just hard to resolve the facts, but conceptually impossible to pass judgment over those who might ski off-piste, overeat, or drink too much wine. Those skiers, eaters, and drinkers may have (and apparently do have) different conceptions of the good life compared to those sitting in judgment. In this sense, it is not just hard, but wrong, to pass judgment.²⁸

In contrast, many ethicists, and even liberal ethicists, favor certain mandates for vaccination.²⁹ Liberal ethicists who embrace mandates must reject the premise that the choice to vaccinate is one of sheer liberty, deserving of liberal deference. Instead, it is permissible to attach consequences to the vaccination decision (via ‘mandates’), whether losing the privilege of flying on airlines or perhaps even losing one’s job. In making such judgments, liberals need not purport to deliver cosmic justice, evaluating the entire worth of a person. Instead, we deliver discrete justice, attaching specific consequences to specific behaviors, as is routine in a legal system. For this reason, unvaccinated persons can lose their jobs, without inquiry into other aspects of their moral worth—for example, whether they also kick puppies or donate to charities. Such policies are routinely implemented with a narrower lens.

Why would such vaccine mandates be permissible, even to liberal ethicists? The rationale is not primarily paternalistic (‘do it because it is good for you’). It is hard to justify the role of the state in protecting people from themselves, since adults of sound mind are typically in the best position know and promote their own interests. Instead, the liberal rationale for a mandate is that vaccination protects other people.³⁰ In the language of economics, we would say that if unvaccinated persons spread infections to others, or deny them access to healthcare, it is an ‘externality’, the classic example of a market failure that demands state intervention.³¹ In the language of political philosophy as developed by John Stuart Mill and Joel Feinberg, those denied healthcare due to pandemic rationing have their interests set back, due to the wrongful behavior of others declining to vaccinate.³² That is to say that the other patients are ‘harmed’ by the unvaccinated.

To be clear, the thesis here is not that unvaccinated people should be blamed for failing to take ‘personal responsibility’. For this same reason, public health scholars support laws that limit smoking in the workplace, because second-hand smoke is harmful to others, even if the smokers should not be shunned and shamed when they seek treatment for chronic obstructive pulmonary disease.³³ The consequences for others raise a quite distinct question.

Morally and legally, we do not hold people accountable for all the consequences of their behaviors, and especially not when the behavior is due to mere negligence. This is another reason why we do not hold a liver cancer patient accountable for drinking too much, even when there is true scarcity of organs for transplantation. Ex ante, if a doctor or a friend were listing all the reasons to moderate your drinking, they might mention the risk of cancer. But they would not typically go further to say that a reason to be moderate is to avoid needing an organ transplant and specifically taking an organ that someone else needed.³⁴ While that may be a factual consequence, it is not one that is foreseeable in the sense that it creates a reason to moderate your drinking in the first place. In the old language of tort law, we could say that the drinking is not a ‘proximate cause’ of the organ scarcity and exacerbating the organ shortage is not a ‘foreseeable risk’ of drinking immoderately. In this context of allocating scarce organs, we may overlook the patient’s history of drinking.

Here again, we see a contrast with vaccination. The consequences of being unvaccinated in a pandemic are much more clearly foreseeable, in part due to the literal immediacy of cause and effect. Specifically, the acute scarcity in the healthcare system is a foreseeable part of being in a pandemic, and that threat is among the key reasons to be vaccinated in a pandemic.³⁵ This fact becomes more obvious every day that the pandemic proceeds.³⁶

Of course, the risk materializes through aggregate action—a single person’s decision to be unvaccinated does not alone create a shortage of ICU beds, but once the risk materializes the given unvaccinated person at the hospital demanding healthcare for COVID-19 is quite proximately threatening to take healthcare that would otherwise be consumed by another patient. And of course, it is not necessary that the unvaccinated person know the specific identity of the person(s) whom they risk harming; we have no problem recognizing that highway deaths are a foreseeable risk of drunk driving or drag racing, even if the driver does not know who specifically they will kill.

Recognizing the existence of foreseeable harms due to rationing, at this point our lens shifts back to the healthcare provider, or other policymaker, who must ration in a pandemic. The familiar, ancient principle of primum non nocere advises to do no harm. A healthcare provider need not judge or punish those seeking care, but if she must ration, she must do so in a way that avoids complicitly harming patients.³⁷

In fact, healthcare providers already recognize the duty to implement policies and practices to protect patients from harm due to behaviors by fellow patients.³⁸ In nursing homes, theft seems to be a real problem, and lax security could make a provider complicit.³⁹ In psychiatric care, patient-to-patient violence, and even homicide, is a known problem to manage.⁴⁰ Beyond thievery and overt violence, there are other mechanisms for patient-to-patient harm. If a hospital knowingly placed a patient or other person in a shared room to smoke cigarettes next to another patient recovering from lung surgery, the hospital would be complicit in the harm caused. Likewise, if a hospital worker left a computer terminal unattended and thereby provided a person’s access to a patient’s medical records, the hospital would be complicit in the eventual breach of privacy.

Accordingly, when advising that physicians may decline to serve unvaccinated patients, the American Medical Association (AMA) has contemplated an interpatient harm mechanism via infection, viz, ‘the risk the patient may pose to other patients in the physician’s practice’.⁴¹ Indeed, even more on point, the AMA seems to recognize that physicians may decline to treat unvaccinated patients ‘if meeting the individual’s medical needs would “seriously compromise” the physician’s ability to provide care needed by their other patients’.⁴² Even outside the pandemic, some pediatricians decline to have unvaccinated children in their waiting rooms, where they could pose a risk to others. In 2016, the American Academy of Pediatrics set a policy in support of such an approach.⁴³

We have, then, the prima facie case for understanding that when a person declines vaccination and then seeks to take scarce healthcare resources to treat the very illness that vaccination would have prevented, displacing others who would have received those healthcare resources, it is a wrongful setback to their interests. That is a harm that rationers should not facilitate. As explained in further detail below, the mechanism for avoiding such complicity in harm is to give patients the same level of care they would have in the counterfactual world in which the unvaccinated persons were not exacerbating the resource scarcity. In what follows, I consider a range of rejoinders to this harm-based approach.

IV. FAIR COMPETITION AND AFFIRMATIVE DUTIES

Here I explore two (technical) objections that press against the harm-based approach. On the one hand, one could deny that there are harms at all when competing for healthcare resources. On the other hand, one could emphasize that harm requires action and that declining to get vaccinated cannot be harmful to all. Both objections are misplaced.

The first objection concerns the nature of the competition for scarce healthcare and seeks to deny that in this context there can be harms, properly conceived. Both Mill and Feinberg recognized that persons have their interests set back ubiquitously—every time we lose a tennis match or a job to some other competitor, we would of course prefer to have won it ourselves.⁴⁴ Even though losing hurts, these ‘fair competitions’ (as Feinberg calls them) do not constitute harms, because they are not wrongful. One might argue that everybody has a prima facie right to healthcare in a society, and thus a right to compete when healthcare resources are scarce. There are no wrongs, in such a situation, and no harms.

Nonetheless, as both Mill and Feinberg recognize, the classic cases of these fair competitions have either of two key indicia: voluntariness or unavoidability.⁴⁵ Some fair competitions, like the tennis match, are entered voluntarily by both parties through something resembling informed consent, which implies the possibility of coming out with fewer points, and thus being the loser. But a pandemic is not a game: vaccinated folks in no way consented to being in a death match with unvaccinated persons. Other fair competitions, like the job search, are unavoidable, part of the very nature of having only one job available for two (or more) applicants. In contrast, in the pandemic case, the acute healthcare shortage is created or exacerbated specifically by one group of persons, the unvaccinated, who dramatically increase their claims on healthcare resources. By assumption in the limited scope of the current argument, if the unvaccinated had simply gotten vaccinated, the death match for this particular person at the hospital door could have been avoided. Thus, for these two reasons, we understand why a victim is harmed when they get punched in the face on the street; we do not creatively reinterpret every such mugging as if it were a fair and voluntary boxing match.

The implicit principle can be made explicit: system healthcare capacity is a collective resource to which each person has a right to their fair share of access in a pandemic, competing with others on neutral rationing criteria, even if their personal interests are instead to receive maximal healthcare. Losing in such a fair rationing competition is just a setback, not a harm, strictly speaking. But when some refuse to take the reasonable step of vaccination, they claim more than their fair share of the limited healthcare resources. In such a case, displacing others would be a wrongful setback to their interests or a harm.

Still, why does vaccination define the moral baseline for fair competition, both as necessary and sufficient? It is necessary, because failing to do so creates foreseeable, proximate harm to others. It is sufficient because it seems to be the only tractable criterion. An alternative, more capacious baseline would be to argue that patients must behave reasonably in the pandemic all-things-considered. Perhaps we would be tempted to deprioritize a patient who attended a crowded wedding, for example? Factually, how many people attended the wedding? Was the patient closely related to the bride or groom, or just attending for fun? How good was the ventilation? When did the patient go to the event and what was the local COVID-19 case rate at the time? Did the patient wear a mask, and if so which type, and how well fitted was it? Resolving these sorts of facts would be impractical for the healthcare rationer, even if she were competent to then evaluate them. As explained above, vaccination is a discrete medical decision, often present in the medical record itself, with clear and consistent public health direction about the reasonableness of undertaking the behavior.

As a second rejoinder, one might invoke the (controversial) distinction between acts versus omissions or misfeasance versus nonfeasance.⁴⁶ Maybe, as some have suggested, persons who decline to get vaccinated do not harm anyone, but merely decline to take ‘an action that presents a benefit to others’ against the exogenous threat of the pandemic.⁴⁷ Even on these terms, one could argue for a ‘reciprocity’ system, where people give a benefit (vaccination) to get a benefit (priority access to healthcare), and would perhaps end up in the same policy place.⁴⁸ But that is not the argument here.

Concededly, the harm principle, like the law of negligence, does not generally require that people take affirmative steps to help other people (ie the principle of beneficence instead).⁴⁹ Yet, to be sure, culpable omissions are often the basis for liability—a surgeon may fail to wash her hands against the exogenous risk of bacteria, a landlord may fail to shovel her steps covered by snow that fell overnight, or a manufacturer may fail to properly warn users of risks when a product is used in certain circumstances. When there are broader policy reasons to impose a duty to act, these are unproblematically understood as cases of harm-doing. The theory of breach is that a person failed to take a precaution she should have taken. Accordingly, reframing the terms somewhat, the contemporary legal approach, reflected in the Restatement Third of Torts, resolves that liability for misfeasance applies to those who create risks.⁵⁰ Thus, a driver who declines to step on her brakes at a red light cannot claim that it was a mere omission, since the very act of driving created the foreseeable risk of collision in the first place.⁵¹ Analogously here, a person who unreasonably declines vaccination creates various risks, including the risk of getting infected and the risk of passing along those infections to others, but also the risk of exacerbating healthcare shortages.⁵² When such vaccine refusers later make a greater claim on the healthcare resources than they otherwise would, the displacing of other patients counts as a harm to them.

Concededly, such talk of affirmative duties again begs questions about the moral baseline, though not perhaps any more than any other moral theory.⁵³ Ultimately, the harm principle rides on more fundamental questions of social policy, to solve collective action problems, specifically here, the taking of optimal precautions to minimize the overall cost of the pandemic.⁵⁴

While more could be said about the act-omission distinction (for whatever it is worth) as applied to the primary decision about whether to vaccinate oneself, the problem disappears altogether when thinking of the role of the healthcare provider, or whomever does rationing of scarce healthcare resources. When undertaking (the act) of rationing, failing to attend to a morally relevant criterion (vaccination) is no different than applying a morally irrelevant criterion (such as race), when deciding who will and will not receive care.⁵⁵ When serving as the gatekeeper, deciding who lives or dies for lack of healthcare, one can hardly demur claiming to do nothing at all.

V. VOLUNTARINESS WITH LIMITED ACCESS AND INFORMATION

These three considerations—true scarcity, a discrete medical choice, and the problem of complicity in harming other persons via healthcare rationing—suggest that allocating healthcare to unvaccinated persons is quite different than the classic cases of RMDH. Nonetheless, one might argue against application of this criterion to a healthcare rationing situation, by suggesting that the decision to decline vaccination is not really a free, informed choice for which one can be held accountable.

As Dan Wikler has said, ‘actions only rarely have all the attributes—informed, voluntary, uncoerced, spontaneous, deliberated, etc.—that, in the ideal case, are preconditions for full personal responsibility’.⁵⁶ This could be yet another reason why clinical ethicists do not hold people who use addictive substances accountable for their behaviors. On the other hand, if actions rarely have all those idealized attributes of voluntariness, we must not routinely require ideal conditions. Bioethics routinely accepts ‘good enough’ consent for a patient to accept medical care or participate in human subjects research.

For the present argument, one particular concern is that some persons may not have reasonable access to the vaccine and thus do not really make a choice to decline it. At least at certain times and places during the COVID-19 pandemic, this concern was substantial, when sheer scarcity made it onerous on some populations to find an available dose. This problem was exacerbated for homebound, disabled, and other persons who are so busy working or providing childcare that it was difficult to find time to get vaccinated. At other times and places, this concern was largely mitigated by the facts that the vaccine had been provided at zero cost out-of-pocket and was widely available, not just via healthcare providers but also at a range of neighborhood sites. Indeed, economic gradients appeared in vaccine uptake, and we can infer that some of this disparity reflects access rather than choice.⁵⁷ Still, later in the pandemic, as vaccination vans roam neighborhoods and cities pay people to get vaccinated, the lack of ‘access’ seems increasingly theoretical, as survey evidence bears out.⁵⁸

A second concern that undermines the moral force of the choice not to vaccinate is that the decision may be made on the basis of misinformation. One could imagine extreme situations where a patient had no relationship with a trusted healthcare provider, was locked in a room with only propagandistic television blaring 24 hours a day, and no access to reliable sources. Such a person could be viewed more as a victim of indoctrination than as a perpetrator of wrongdoing. This problem may well cut along political lines in the future, just as it did under COVID-19.⁵⁹ For many, the choice not to vaccinate is the choice to buy into an ideology and identity that prizes a contrarian perspective that disvalues the risk to vulnerable persons and then curates media to reinforce that belief, while avoiding or downplaying contrary evidence.

Aside from politics and ideology, one might also be concerned with educational disparities and lack of trust in the healthcare system, driven by historic exclusion and exploitation.⁶⁰ In this light, one might worry that vaccination priority would worsen disparities. Nonetheless, as Persad and Largent review the empirical evidence later in the COVID-19 pandemic, it becomes clear that ‘Americans who are members of racial and ethnic minority groups report higher than-average vaccination rates, as do women and people with serious medical conditions’.⁶¹ Thus, the racial disparity story is not borne out. Moreover, in constructing a fair rationing policy, one must remember that disadvantaged persons could be on either side of the ledger, either as the unvaccinated patient making disproportionate claims on healthcare resources or as the other patient who would have gotten the care instead.⁶²

These two concerns about voluntariness have facial validity, but they are contingent and defeasible. In a future pandemic, the facts will need to be evaluated before assessing that all persons can be held accountable if they decline to vaccinate, but it seems dogmatic to suggest that a reasonable degree of access and choice could never be attained. To be sure, at any point on these gradients of information and access, policy makers should consider additional efforts to maximize access and minimize disinformation.⁶³ For example, vaccinations can be delivered door-to-door. The government might also do a better job of communicating to diverse sectors of society, for example, using spokespersons that have broader political appeal. In addition, prior to implementation, a policy that vaccination status will be considered in rationing decisions should be announced and disseminated broadly, which may (or may not) further increase the rate of vaccinations or at least add another informational opportunity.⁶⁴ Before such a policy should be implemented, a more systematic approach to recording vaccine status might also be required.⁶⁵

Fundamentally, however, there is a normative mismatch between this concern about voluntariness and the problem at issue here. When scholars say, ‘it may be difficult to know the true reasons a patient is unvaccinated’, it is true enough.⁶⁶ But also completely inapposite, because for preventing one person harming another, we employ an objective, rather than subjective standard.⁶⁷ Accordingly, the question is not what the person believed or intended when she refused vaccination, but what a reasonable person would have done. We do not tolerate one person harming another merely because they had good intentions or were suffering from confusion.

The analysis may be different if the argument were about punishing the unvaccinated, rather than avoiding complicity in their harm-doing.⁶⁸ In the COVID-19 pandemic, we might then imagine and worry about counterfactuals where Fox News had made different programming choices, or going further back, that the Tuskegee Airmen Study had never happened, such that citizens for whom those facts are salient to their trust in government or healthcare institutions would now make different vaccination decisions. Similarly in other cases of interpersonal harm, we may wish that cult leaders or mob leaders had never influenced their followers, who go on to harm others. We might wish that some people with psychosocial disorders had had better access to mental healthcare or addiction recovery services, prior to harming others. But none of these counterfactuals, in any of these cases, would undermine the argument for acting in real-time to prevent the doing of harm to others. Even more, none of these counterfactuals would excuse healthcare rationers from becoming complicit in doing their harms to others. When someone goes to rob a bank, we do not hand them the keys to the safe or give them a gun to shoot the bank teller, just because they have some confused ideology or false beliefs or lacked good access to something they needed. And the mere fact that there is some other wrong-doer in the background, using the bankrobber as a mere puppet for her own nefarious ends, does not change the analysis about complicity or intervention in harm-doing.

Recall that many ethicists still favor vaccination mandates, even for profound consequences like losing one’s livelihood. This embrace of mandates suggests on the one hand that vaccination is not the sort of thing where voluntariness is all that worrisome, and moreover, we are willing to impose profound negative consequences (such as loss of a job) notwithstanding these sorts of concerns about access and information. This, again, is because being unvaccinated foreseeably risks harms to other persons. In this context, the voluntariness concern loses its force.

Similarly, we might worry that nicotine addiction makes a patient’s choice to smoke another pack of cigarettes less than fully voluntary. But that does not mean that the hospital should be complicit in facilitating the smoker to pollute the air of the patient recovering from a lung transplant. Likewise, perhaps a person is mentally ill, suffering from some sort of obsessive disorder, which compels him to review and publicize a patient’s medical records. Here likewise, the lack of voluntariness does not reduce the healthcare provider’s duty to avoid complicity in harm.

Still, even in the context of harm to others, we can implement rare exemptions for the vaccination priority in healthcare rationing, clearly some medical, and perhaps others religious or based on the foregoing considerations about access and information. One key exemption would be for minors or others who are incompetent to decide for themselves, but whose parents or other surrogates refused vaccination. Perhaps the best conception is to make priority for those vaccinated a default rule subject to these exemptions only in special cases, which are well-defined and procedurally fair. Any such rule will be imprecise, but not nearly as crude as the alternative rationing policy that systematically ignores vaccination status altogether and thereby wrongfully perpetrates consequences of being unvaccinated on those who are vaccinated.

VI. PROPORTIONALITY AND UTILITY

It bears emphasis that the analysis so far only applies to cases of acute healthcare scarcity, where rationing at the point of care is necessary. This situation generally does not apply outside of pandemics, and even then only at times when healthcare demand is highest. Moreover, this thesis only applies to unvaccinated patients who are seeking care specifically for the illness for which safe and effective vaccines are available (though it may protect other patients’ access to healthcare for a range of ailments that could be impinged by acute healthcare shortages).

Even in this narrow scope, one must determine how much to deprioritize unvaccinated patients, especially given other broadly utilitarian rationing criteria like medical efficacy and efficient use of limited resources, and still other criteria such as priority for the worst-off. According to the harm prevention argument herein, when performing that rationing function, healthcare providers must avoid being complicit in externalizing the foreseeable consequences of being unvaccinated upon others. The rule then is to give the patients the same level of care that they would have in the counterfactual world in which there is a pandemic, but the unvaccinated persons were not creating or exacerbating the resource scarcity due to their decision to be unvaccinated. This net equivalence counterfactual lodestar is similar to that which has been argued for giving healthcare workers priority access to healthcare.⁶⁹

While counterfactual reasoning is essential to any causal analysis, we cannot know with certainty the details of the counterfactual. Nonetheless, review of metrics, including the proportion of patients seeking treatment for COVID-19 who are unvaccinated compared to those vaccinated, gives a clear sense of the number of beds being occupied on the basis of the choice to be unvaccinated. For the sake of illustration, one can imagine a case where the unvaccinated patients are taking 8 out of the 10 ICU beds for pandemic-related care, but we estimate if fully vaccinated, those same persons would only consume one ICU bed. This suggests that the unvaccinated patients have a full claim on the one bed, but that the other patients should have priority access to the remainder of the beds at least to the point that they are not at risk of serious harm for lack of such a bed. This analysis is complicated by the idea that one goal of this proposal is to protect healthcare access even for healthcare needs that are not related to the pandemic (eg an ICU bed for someone who suffers from a drowning). Both vaccinated and unvaccinated patients have a full claim on such healthcare resources, and thus both receive protection under the harm principle.

These sorts of back-of-the-envelope analyses should be discomforting, when life and death are at stake. In some cases, like those in classic RMDH, if a moral determination is too hard, then it is best simply avoided. Here a healthcare rationer cannot simply avoid the problem by ignoring vaccination status, anymore than they can just ignore other rationing criteria like maximizing benefits, which can be sometimes hard to predict.

At the very least, this analysis suggests that the harm prevention principle should take priority over largely arbitrary criteria such as lottery-based allocations, and arguably other rationing tools like ability-to-pay or rationing-through-inconvenience, which may have systemic value, even if poorly calibrated.⁷⁰ ‘First-come first-served’ is a ubiquitous rationing criterion that is similarly arbitrary, but it also implicates a broader question about whether rationing decisions must be continually updated to maximize criterion achievement, or whether interests in healthcare resources become vested upon formation of a doctor–patient relationship or upon receipt of another resource. For example, if a patient has been properly allocated a ventilator, but is no longer the best use of the limited resource (under any rationing criterion), should she be extubated to allow reallocation?⁷¹ Such hard questions are not peculiar to any particular rationing criterion, including vaccine priority, so I set them aside here.

The present argument becomes more difficult if the default criteria for rationing healthcare instead reflect utilitarian concerns, specifically who would get the most benefit from a healthcare resource on the margin, compared to not getting the healthcare resource. Accordingly, the present proposal could be counter-utilitarian as applied, giving healthcare to another patient when it would do more good for the unvaccinated person, reducing overall health and social welfare.⁷² Strikingly, beyond this case of pandemic vaccination, if we supposed that utility outweighed the harm principle, then we ought to allow some to harm others as long as they get more utility than the harm they cause, which presents something of a reductio ad absurdum for simplistic utilitarianism. Of course, one classic way out of this dilemma is to suppose that equilibrium effects will cause a policy of harm prevention to also turn out to be utility-maximizing, in the spirit of rule utilitarianism. For example, a policy of deprioritizing those who refuse vaccination may cause more people to get vaccinated, thereby avoiding such extreme scarcity of healthcare and saving more lives in the end. This is an empirical claim, of course, and while plausible cannot be defended conclusively here.

In any case, the need to maximize medical benefits is particularly important and arguably must be balanced against this need to avoid being complicit in doing harm.⁷³ One plausible approach is to first rank patients by likelihood of benefit, before applying any other criteria.⁷⁴ Alternatively, there are other proposals for allocation mechanisms that can incorporate multiple values.⁷⁵ The present thesis is just that, in certain circumstances, vaccination status is a valid, obligatory criterion among others.

VII. SOLIDARITY AND COGNITIVE DISSONANCE

In a systematic review of the literature on rationing criteria, scholars have commented that, ‘As the COVID-19 outbreak is potentially life-threatening, the need to stick to the universal rules of ethics – justice, benevolence, and distributive justice are of paramount importance’.⁷⁶ The foregoing arguments could be wrongly read as seeking to set aside fundamental moral principles in a time of pandemic crises. Instead, these arguments have sought to take the large principles reflected in the consensus view against RMDH seriously on their own merits, but find them to yield surprisingly different conclusions in the rare and peculiar case of pandemic vaccination.

Nonetheless, the solidaristic consensus against blaming patients (defined as ‘RMDH’ above) may be so strong that readers may seek to maintain the consensus universally, against the force of these harm-based arguments.⁷⁷ The resistance may be almost instinctual, functioning at the level of identity. As Dan Wikler has written, it is important to maintain ‘…the very useful and virtuous first instinct of the doctor or nurse, that of sympathy and care for the suffering’.⁷⁸ Zooming out, Wikler suggests that, ‘The same point can be made for societies as a whole: it is not salutary for people to become used to withholding sympathy for sick fellow-citizens…’.⁷⁹ From the eyes of patients, White and Lo strike a similar note when they write that, ‘Categorically excluding patients [from receiving healthcare] will make many feel that their lives are “not worth saving,” which may lead to perceptions of discrimination’.⁸⁰

As a threshold question, it is not particularly clear why this vaccination criterion for rationing implicates such concerns any more than any other rationing criterion, where someone will be denied care. Lines will be drawn, and care will be denied, regardless of the criterion.

To be sure, we should indeed feel sympathy for and solidarity with all those needing healthcare. It is, concededly, divisive and perhaps even corrosive, to carve a line between fellow humans, all of whom desperately seek healthcare. Clearly, in some ideal sense, the most virtuous course would be to give the unvaccinated persons a dispensation, an exercise of grace, or an indulgence in the spirit of forgiveness. Generosity is a moral virtue after all, and as Gheaus echoes Socrates, ‘only a morally good life can be a good life’.⁸¹

As courts have often said, however, ‘one must be just, before being generous’.⁸² It is not generous to help one by harming another. In this sense, the impulse to help all patients is not wrong; it is just overridden by a stronger moral command in times of acute scarcity. It bears emphasis that the ancient principle of primum non nocere is aptly put because compared to beneficence, the overriding first duty is to do no harm.

While I think correct in this case, this simplistic notion of value-ranking, where one value trumps another, may be unsatisfactory. Indeed, the moral dilemma is real, and both sides should be felt, even if hard. Such a conflict of values creates cognitive dissonance, which feels bad.⁸³ As Dan Brock has explained, the field of bioethics has from its inception had a dual role, on the one hand, resolving issues arising in clinical medicine, and on the other hand defending ‘a general moral right to healthcare’.⁸⁴ However, ‘the focus on establishing a right to health care has contributed to a failure to address difficult issues in developing ethical standards for equitably prioritizing limited resources in health care’, the present inquiry.⁸⁵ Indeed, even the word ‘rationing’ has been avoided by some scholars, playing into the politicians’ game of pretending ‘that rationing did not occur, was not necessary, was politically dangerous, and would be morally wrong if it did occur’.⁸⁶ This is one way to avoid the cognitive dissonance, but it is not right.

The best answer to these concerns may be to emphasize the limited scope of the present argument and thereby to suggest that it will not override broader commitments to solidarity as applied in millions of everyday cases. In classic cases, the lack of acute scarcity and intractability of factual discernment are bright lines that distinguish vaccination priority in a pandemic and thus avert worries of a slippery slope. At the same time, I also want to normalize the current argument, invoking common-sense examples (like a patient smoking in a hospital or viewing another’s medical records) where healthcare providers already act to protect patients from others harming them.

Denial of healthcare should be a last resort, but unfortunately during a pandemic, such a grim reality may become unavoidable. If that conclusion seems strange and unfamiliar, it is because the factual circumstances are thankfully exceedingly rare. We do not have the moral practice or habit formation that supports these sorts of decisions.⁸⁷ That is why it is useful to make these sorts of analyses from the luxury of time between one instance of pandemic scarcity and another, which will predictably arise in the future.

VIII. CONCLUSIONS

Concededly, one could arrive at the same thesis via other routes, such as luck egalitarianism or sheer utilitarianism, with certain empirical assumptions and broader implications. Instead, I have tried to construct this argument in a way that does no violence to the general consensus against RMDH and thereby maintain as many of our status quo commitments as possible. I view the harm principle as something of a minimum, where there may be more overlapping consensus, and thus a potential firmer footing.

These considerations apply in very specific circumstances, which will not often arise outside certain pandemic situations. These circumstances are: a vaccine is widely available for free and proven safe and effective, where it is foreseeable that actual rationing decisions will need to be made, where in fact that extreme scarcity arises, where an unvaccinated patient seeks care that would have been unnecessary if vaccinated, where vaccination status can be reliably confirmed, and where exemptions do not apply.

In those cases, to prevent harm to others, it is appropriate to consider vaccination as a criterion for the allocation of healthcare resources, and indeed wrong to not do so. Ultimately after all, somebody will be denied healthcare. The question is, who?