Ethical heuristics for pandemic allocation of ventilators across hospitals

In response to the COVID-19 pandemic philosophers and governments have proposed scarce resource allocation guidelines. Their purpose is to advise healthcare professionals on how to ethically allocate scarce medical resources. One challenging feature of the pandemic has been the large numbers of patients needing mechanical ventilatory support. Guidelines have paradigmatically focused on the question of what doctors should do if they have fewer ventilators than patients who need respiratory support: which patient should get the ventilator? There is, however, an important higher level allocation problem. Namely, how are we to ethically distribute newly obtained ventilators across hospitals: which hospital should get the ventilator(s)? In this paper, we identify a set of principles for allocating newly obtained ventilators across hospitals. We focus particularly on low and middle income countries, who frequently have limited pre-existing intensive care capacity, and have needed to source additional ventilators. We first provide some background. Second, we argue that the main population healthcare aim during the COVID-19 pandemic should be to save the most lives. Next, we assess a series of potential heuristics or principles that could be used to guide allocation: allocation to the most densely populated cities, random allocation, allocation based on the ratio of patients to ICU personnel, prioritisation in terms of intrahospital mortality, prioritisation of younger populations, and prioritisation in terms of population mortality. We conclude by providing a plausible ranking of the principles, while noting a number of epistemological challenges, in terms of how they best further the aim of increasing the probability of saving the most lives.


PALACIOS-GONZÁLEZ ET AL.
COVID-19 appears to be lower than that of SARS (9.5%) and Middle (0.1%)." 2 And more recently Yang Cao et al. have concluded that the 3 On March 11, 2019 the WHO declared COVID-19 to be a pandemic. So far it has spread to 219 countries and territories, and it has killed more than two million people. 5 It was predicted that if left unchecked COVID-19 could have infected 60% of the global population. 6 In the UK, for example, the worst case scenario predicted that 80% of the population could have become infected.
The tackling of COVID-19 raises many ethical questions.
Among them, the question that has received the most attention is how to ethically allocate scarce medical resources, specifically how to allocate ventilators. 8 Ventilators are centre stage 9 on the ethical discussion because international experience showed that even well-resourced healthcare systems had serious problems coping with a large influx of COVID-19 patients requiring them. 10 In low and middle income countries (LMICs) health systems experienced intense pressure (Box 1).Unlike some other medical resources (e.g. medication, personal protective equipment), ventilators are highly precise instruments which cannot be easily mass produced in a short time. And similar to other medical resources, ventilators are tightly regulated. This means that it does not matter how fast companies produce new ventilators, the demand created because of COVID-19 will outpace global supply in the short and medium term. 11 All this means that for the time being hospitals around the world will have to work with existing capacity while waiting for new ventilators to trickle down the healthcare system. And this trickle down can also be affected by disruption to global supply chains.
The ethical discussion about how to allocate ventilators during the COVID-19 pandemic has largely focused on how they ought to be allocated between individual patients. Whilst this is an important question, there is an important gap in this discussion; current ethical debates do not consider how governments in LMICs should allocate newly obtained ventilators across hospitals. Here 'obtained' must be understood as governments getting hold of more ventilators from manufacturers, or national or international organisations, or from something akin to a national stockpile. The best known example is the US Strategic National Stockpile, whose role is to supplement state and local authorities' supplies during a public health emergency, and it includes ventilators. 19 Much of the current ethical debate about allocating ventilators follows what we can call the intra-hospital model assumption of allocation. This is, people theorise about the allocation of scarce medical resources while assuming: i) that all the available resources to treat certain disease are already in a hospital, ii) that patients arrive to such a hospital, and iii) that after patients arrive to such a hospital the allocation process takes place.
The aim of this paper is to close the aforementioned gap by exploring a series of principles for governments to allocate newly ob-   13 Interestingly, the Mexican government never publicly said how many ventilators they thought they would need in order to adequately deal with the pandemic, even when they publicly opted for a mitigation strategy rather than an elimination one.
However, foreseeing that the amount of available ventilators would not be enough for the country the government set out to buy more. They did so not only because of the comparatively small number of ventilators to population that they had, but also because at view, from a utilitarian position, we might appeal to the dictum attributed to Bentham that 'everybody to count for one, nobody for more than one'. If everybody counts for one, and we all have the same value, then it seems intuitive to accept that the more people we can save the better. This insight is captured by Derek Parfit when he writes, "Why do we save the larger number? Because we do give equal weight to saving each. Each counts for one. That is why more count for more." 23 Non-utilitarians can accept this principle too, but a coin between one person and any number on the other side, giving each person an equal chance, we would behave no differently than if it were a contest between one and one, where equipoise can be resolved by the coin toss. If the presence of each additional person would make no difference, this seems to deny the equal significance of each person".
In working out how to best meet this fundamental goal, we must consider the implications of different allocation decisions across time. To illustrate the point in the context of individual allocation decisions, in such contexts we may have to decide whether to allocate a ventilator to patient X for two weeks, or to patients Y and Z for one week each. In thinking about allocation across hospitals, we may have to consider whether to allocate a ventilator to a hospital that is experiencing a surge in cases now, or one that may experience a more serious surge later. Of course, whilst saving lives is the fundamental goal of ventilator provision, there are plausibly other moral considerations that should feature in our allocation decisions. In particular, it might be argued that allocation decisions should also be grounded by principles of equality and fairness, and that they ought not to instantiate morally problematic forms of discrimination. 25 Whilst we are sympathetic to this claim, we nonetheless believe that what we have identified as the fundamental goal of ventilator provision should take some degree of precedence over other moral considerations in allocation decisions. 26 When thinking about the allocation of ventilators to individual patients, if we accept that we should use ventilators in order to maximise the numbers of lives saved then it follows that they should be allocated to those patients with the highest chance of surviving if treated, and who will need the ventilators for the least amount of who will need it for 1 month, or we could allocate the same ventilator the latter. However, for reasons that we shall explain later, the application of the save the most lives principle is more complex in the paper we will accept that one of our main aims during the COVID-19 pandemic should be to save the most lives, and thus the allocation principle(s) for newly obtained ventilators across hospitals should, perhaps within certain limits, seek to further this aim.
Before moving forward, some might still resist our stance and argue that considerations of equality/fairness should take priority over the number of lives saved. We accept that there is a radical tension between the save the most lives principle and other principles grounded by considerations of equality/fairness, and that in practice any triaging policy has to make a call on how to weigh these two competing stances. Even when here we cannot adequately address this fundamental issue, our discussion in this paper will add further dimensions to this general debate. It will do so by showing that how we choose to operationalise the save the most lives principle in the allocation of ventilators in LMICs has implications for the precise allocating ventilators to hospitals that have comparatively low mortality ratios might unfairly benefit comparatively well-off communities. We might avoid this outcome by adopting a different principle, but any other principle we outline will introduce different forms of inequality/unfairness. In the end, ranking the principles that we next explore will be guided by how well we think they operationalise the goal of saving the most lives; however, this ranking will also be influenced by views about which sorts of inequality/unfairness we are willing to accept in the name of saving the most lives. densely populated cities usually have the largest number of hospitals, which comparatively speaking are well equipped in terms of pre-existing resources and personnel (for example large tertiary hospitals). This would potentially increase the chance that any individual ventilator is able to save a life. A ventilator would go to waste, for example, if we allocated it to a hospital where there is not sufficient medical expertise for operating it. Second, allocating ventilators to hospitals within the most densely populated cities has the theoretical advantage that it does not matter to which specific hospital we allocate it. That is because we can transfer a patient from a hospital where there are not enough resources to another hospital with available resources. The hospitals would be geographically proximate. By contrast, if ventilators were allocated to Oaxaca, for example, hospitals are so distant from each other, and the roads so bad, that transfer of critically ill patients between hospitals is likely to be impossible.
Although intuitively appealing, automatically allocating newly obtained scarce medical resources to hospitals within the most is not always straightforward in some of the most densely populated nationwide, the ambulance infrastructure is not there to safely demic communications among hospitals, and of hospitals with the because of poor coordination, and this has been a common feature throughout the pandemic. 28 Second, there are small(er) cities where their hospitals interact efficiently on a regular basis. In the case of Mexico, for example, Guadalajara seems to do better in this regard than Mexico City. This means that they could transfer acutely ill patients from hospitals without resources to hospitals with resources. All the former discussion about transferring patients presupposes that hospitals in the most densely populated cities and those of the small cities are operating at capacity or close to capacity, but there is no (triaged) waiting list for accessing a ventilator. Once all hospitals have (triaged) waiting lists for accessing ventilators the point about transferring patients becomes moot for the most part, other things being equal. Why? Because (triaged) high priority patients in the waiting list should receive the new ventilator(s). However, it is also true that hospitals in small cities could suffer from the same communication and logistical problems as those in densely populated cities. We can thus conclude that when searching for an allocation principle for newly obtained ventilators we should not assume that acutely ill patients can be transferred across hospitals safely.
Third, we require at least one additional principle to determine how many new ventilators should be allocated to each highly allocated to Ecatepec, city of 1.6 million people, in contrast with ciple we are left out in the dark about how to allocate the new ventilators. This issue is not per se an objection to the principle, but it shows its limitations. densely populated cities have the largest number of hospitalsand that they are well equipped in terms of resources and personnel -it does not automatically follow that more lives would be saved if they were to receive the newly obtained scarce medical resources first. Because, other things being equal, the ventilators would save the same number of people in a highly densely populated city as in a less densely populated city. A single ventilator would be able to treat, at a time, one patient in Mexico City or one patient in Cancún. And the same is true even if the resources were to be spread across very small cities. We can thus conclude that the population density principle probably would not lead to saving the most lives. 29 Before moving to the next section, it could be objected that we have not taken into account that the basic reproduction number 30 is likely to be greater in densely populated cities than in less densely populated cities, and that thus allocating more ventilators to the former ones would save more lives than allocating them to the latter ones. The quick answer to this question is that we are presupposing that hospitals in both types of cities are working at capacity, or close to capacity, and thus that even when they could have different basic reproduction numbers the situation at the hospitals is the same. It is true that after the initial allocation of new ventilators certain cities might better handle the disease, and their hospitals could no longer be working at capacity, or close to capacity. If this were to be so then we might reallocate ventilators from that city to other cities. A second objection is that higher population density cities will have more patients waiting for a ventilator, and higher priority patients on the (triaged) waiting list. However, this will not necessarily be the case. If less densely populated cities have a lower pre-existing intensive care capacity, or have a worse outbreak, they could have more patients waiting for a ventilator (with a better prognosis) than those in the densely populated city. One reason why randomness as a moral principle is attractive, according to John Broome, is that "If a good or bad cannot be distributed equally, it sometimes seems a good idea at least to distribute it randomly. Randomness appears to be a way of bringing some fairness into an inherently unfair situation". 31 In this particular case, fairness seems to be achieved in that each hospital has the same chance of getting the ventilator(s). And furthermore, this particular principle of allocation easily resists corruption when carried out transparently. One serious political concern about allocation of ventilators across regions and hospitals, is that such distribution might be susceptible to political influence. Politicians elected from a particular city or region might seek which is a key battleground state during US presidential elections, was perceived as receiving preferential treatment from the former US administration in terms of resource allocation. 32 A random process of allocation would avoid such influence and avoid concerns about biased allocation.
Suppose that we have one new ventilator and we are faced with two hospitals, which serve the same type of population, that are working at capacity while patients continue to arrive. In this particular case, other things being equal, the just way to allocate the ventilator would be randomly allocate it (e.g. via a lottery). If other things are indeed equal then we should resort to a random allocation principle. However, it is a mistake to assume that the default position when comparing two (or more) hospitals working at capacity is that things are equal. In order to understand why we need to pay attention to how ventilators work.
A ventilator is not like a magical life preserver that every time we throw it at a group of drowning people it catches someone and does not let her go until she is safe at the beach. Ventilators are just one tool, among many, within ICUs that need to be carefully monitored and managed by doctors, nurses and therapists. 33 When properly utilised, they increase the probability of a patient surviving vere respiratory failure they may increase the chance of surviving from 5% to 50%. However, the magnitude of this increase in chance is not fixed. Thus, when deliberating about how to allocate newly obtained ventilators we should consider how their use increases the probability of saving the most lives. It might appear that this change of focus from i) saving the most lives to ii) increasing the probability of saving the most lives is merely semantic, but it is not for two reasons.
Second, because it allows us to better identify how things can be different in two hospitals working at capacity or close to capacity.
There is evidence that as the ratio of critically ill patients to ICU healthcare personnel increases the probability of patient survival decreases. It does so because clinical attention is spread across several critically ill patients that need to be carefully monitored, among other reasons. Imagine the two following scenarios: a hospital has six ICU beds and two ICU nurses per working shift, and another hospital also has six ICU beds but three ICU nurses per working shift.
This means that we have a patient nurse ratio of 6:2 and 6:3.

Allocating a single ventilator randomly between these hospitals
It is obvious that if we want to increase the probability of saving the scenario; and thus reject a random allocation principle for this case.
This leads us to conclude that if things are not equal in terms of hospital equipment, personnel, and number of patients treated then we should first pay attention to how allocating newly obtained ventilators increases or decreases the probability of saving the most lives.
We discuss this further in the next section.  While theoretically attractive, this principle is pragmatically hard to implement in a reliable way. | One of the distinctive features of the COVID-19 pandemic is its disproportionate impact on older members of the population: the death rate increases with age. Thus, if we want to improve the probability of saving the most lives then we should allocate newly obtained ventilators to those hospitals that not only have the necessary extra resources and lowest patient to ICU personnel ration, but are also located in places where the population is comparatively younger. Let us call this the Population Age Principle.
Population Age Principle: allocate newly obtained ventilators to hospitals serving a younger population.
Suppose that the mean age of the population being served by hospital A is 50 years old whereas the mean age of the population being suming that both hospitals have exceeded their intensive care capacity), we should allocate the extra ventilator to hospital A. Doing so increases the probability of saving the most lives. It could be objected that allocating ventilators to hospitals in this way is a form of ageism.
The claim might be that we are unjustly discriminating against those populations with the highest number of older people because we consider that their lives are of less value. However, the population age principle is not inherently tied to a specific age, but rather it is responsive to evidence about specific diseases and survival rates. It would mean, for example, that in a different pandemic, that affected younger patients more than older patients (e.g. as occurred in the 1918 influenza pandemic), other things being equal, we would allocate new obtained ventilators to hospitals that are located in cities where the mean age is older. It is important to highlight that once we had allocated newly obtained ventilators to those particular hospitals internal resource allocation protocols would still take place.
Whereas the population age principle furthers our aim of increasing the probability of saving the most lives there are at least two pragmatic considerations that need to be taken into account ments in some LMICS do not have up to date data on the age-range of the population that a certain hospital serves during the pandemic.
This can be so because the government does not collect or curate the data properly, or because the pandemic causes migration patexample, earlier in the year India saw mass internal migration due to lockdown measures. Second, there could be shielding measures in place to protect the elderly from coming into contact with the virus. mortality rate is not solely linked to biology, but to social determinants of health. Such populations suffer higher rates of pre-existing poor health, suffer higher rates of malnutrition, are more likely to live in deprived areas, are more likely to live in inadequate and overcrowded conditions, and are more exposed to pollutants, all which could influence mortality from  Therefore, if we were to follow a general population mortality principle, other things being equal, then it is quite possible that the newly obtained ventilators should be allocated away from the hospitals that serve these communities.
Of course, this troublesome implication of following a general population mortality principle in the allocation of ventilators will only arise in certain countries. There are high income countries, for example Japan or Norway, and low and middle income countries, for injustices, for example indigenous communities. It might seem that the best course of action in such circumstances is to allocate the new ventilators to the hospitals serving those communities. Doing so could end up in three different scenarios, depending on the hospifew material resources but an adequate numbers of personnel. In this scenario we end with the same allocation result as if we were following the Patient to ICU Personnel Ratio Principle. Second, we could have a hospital that is completely stretched out in terms of personnel (or other material resources) and where receiving additional ventilators would not benefit any patient. It is obvious that if this were to be the case then this hospital should not receive the new ventilators (unless it can be assigned additional personnel and other material resources) and they should be sent to another hospibe cases where the allocation of new ventilators would not further the aim of increasing the probability of saving the most lives, but nonetheless would increase the probability of saving lives. Let us remember two things: i) these three scenarios only obtain if we act against the General Population Mortality Principle where it disadvantages communities affected by health inequalities and structural injustices, and ii) that once the new ventilators have been allocated internal triage policies would still apply. In conclusion, if we are at all attracted to increasing the probability of saving the most lives, we will at least need to consider these principles in the light of available local information. In this way, practical ethics is context dependent.