We wholeheartedly agree with Schmidt and colleagues’ efforts to promote equity in intensive care unit (ICU) triage. We also take issue with their characterisation of the New Jersey (NJ) allocation framework for ICU beds and ventilators, which is modelled after the multi-principle allocation framework we developed early in the pandemic. They characterise it as a two-criterion allocation framework and claim—without evidence—that it will ‘compound disadvantage for black patients’. However, the NJ triage framework—like the model allocation policy we developed—actually contains four allocation criteria: the two criteria that the authors mentioned (chances for survival and near-term prognosis) and two criteria that they failed to mention which we included to promote equity: giving priority to frontline essential workers and giving priority to younger patients. These omissions are problematic both for reasons of factual accuracy and because the two criteria they failed to acknowledge would likely mitigate rather than exacerbate racial disparities during triage.
- allocation of health care resources
- distributive justice
- public health ethics
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
We agree with Schmidt and colleagues that mitigating racial and ethnic inequities in health outcomes is an important goal when allocating scarce lifesaving treatments during the COVID-19 pandemic.1 Their efforts to advocate for equitable allocation frameworks have been important in reshaping the debate about fairness in triage during the COVID-19 pandemic.
However, we take issue with the authors’ implication that the New Jersey (NJ) triage framework is inattentive to equity concerns. This issue caught our attention because the NJ framework is modelled on the multi-principle allocation framework we developed early in the pandemic.2 Contrary to the author’s characterisation of it, our framework is designed to both promote population health outcomes and promote equity.
The authors characterise the NJ framework as containing only two allocation criteria: chances for survival and near-term prognosis; they wrote: ‘NJ’s Directive combines the SOFA score with another score assessing near-term survival to produce an overall raw score. This raw score is then used to assign patients to priority groups’.
However, the NJ triage framework—like the model allocation policy we developed—actually contains four allocation criteria: the two criteria that the authors mentioned (chances for survival and near-term prognosis) and two equity-focused criteria that they failed to mention: giving priority to frontline essential workers and giving priority to the young. These omissions are problematic both for reasons of factual accuracy and because the two criteria they failed to acknowledge would likely mitigate disparities during triage, as described below. Table 1 contains a summary of the four criteria in the NJ framework.
Disparity mitigation from prioritising frontline essential workers
We advocated giving heightened treatment priority not only to healthcare workers, but also to ‘individuals who perform tasks that are vital to the public health response’ and those who ‘play a critical role in maintaining societal order’.3 This language was intended to describe the individuals who have come to be known as frontline essential workers, who assume heightened risk in jobs vital to safeguarding society during the pandemic. Frontline essential workers include not only healthcare workers but also lower-paid workers who are likely to experience multiple social and economic disparities,4 such as grocery store clerks, bus drivers, agricultural workers and custodial workers. Incorporating this criterion into triage would likely create a substantial disparity-mitigating effect because frontline essential workers comprise a very large group who are disproportionately persons of colour.5 Moreover, recent studies have found that frontline essential workers who contract and die from COVID-19 are disproportionately from racial and ethnic minority groups.6 7
Disparity mitigation from prioritising younger patients
We advocated giving heightened priority to younger patients by using age as a tiebreaker between patients who have similar triage scores but where significant differences in age existed. For example, if a 25-year-old patient and an 85-year-old patient presented with similar triage scores, we argue that priority for the scarce intensive care unit (ICU) bed/ventilator should go to the younger patient. Although some authors have mistakenly claimed that prioritising younger individuals expresses animus toward older adults or devalues the contributions that older adults make to society,8 properly understood, the ethical justification is to promote fair opportunity to live through life’s stages by prioritising the worst off.9 10 In this sense, dying young is a severe form of disadvantage because young individuals will have died with fewer opportunities to formulate and carry out their life plans. Therefore, all else being equal, younger patients have a stronger claim to absolutely scarce life-saving resources than older individuals, who have had more opportunity to live through life’s stages.
A consequence of giving some priority to younger individuals during triage is that it would counteract racial and ethnic disparities in COVID-19 outcomes. The reason is that, on average, racial and ethnic minorities are being hospitalised and dying at significantly younger ages than white patients. For example, whereas only 13% of deaths among white patients occurred among persons less than 65 years old, 30% of deaths among black patients and 35% of deaths among Latino patients have occurred in this age group.11 Even more striking is that 78% of deaths among individuals younger than 21 years have occurred among racial/ethnic minorities, despite the fact that they comprise a much smaller proportion of that age group.12 In addition to counteracting racial disparities, giving some priority to young patients would likely offset disadvantages experienced by individuals with life-shortening disabilities who are more likely to die from COVID-19 at an earlier age (eg, patients with Down syndrome, cystic fibrosis or various paediatric malignancies).
The need for evidence about the real-world impact of triage policies
Schmidt and colleagues claim—without evidence—that the NJ framework would ‘compound disadvantage for black patients’. Although it is plausible—but uncertain13—that using the two-criterion framework they inaccurately described as the NJ framework may worsen disparities, the actual four-criterion framework we developed is unlikely to worsen disparities and would probably mitigate them. In a study involving a diverse cohort of more than 1000 patients admitted during the COVID-19 pandemic surge, Gershengorn and colleagues found that a two-criterion triage framework based on chances of survival and near-term prognosis did not produce racial inequities in triage between black, Latinx and white patients.13 Notably, the two-criterion framework they studied did not contain the two equity-focused criteria included in the NJ framework. This study of a racially and ethnically diverse cohort of COVID-19 inpatients at risk for ventilator triage demonstrates the value of and the need for empirical studies to assess the real-world impact of policies on equity. Based on these findings, it seems more likely that an allocation framework like NJ’s that contains two efficiency-focused criteria plus two additional equity-focused criteria would lessen disparities rather than increase them.
It is important to acknowledge that there is uncertainty about the real-world effects of existing triage frameworks on both efficiency (ie, the number of lives saved) and equity (ie, the fair distribution of those benefits across patients). Robust empirical data about the probable impact of various triage frameworks could meaningfully inform policy debates about allocation criteria for scarce life-saving resources. One promising approach to evidence generation is to use modelling methods applied to cohorts of actual patients admitted during the pandemic to simulate the probable impact of various triage frameworks on key outcomes of interest, including the number of lives saved and whether triage exacerbates or mitigates racial inequities. Ideally, such data would allow policymakers and other community stakeholders to make informed decisions around triage frameworks.
More potent strategies are needed to mitigate racial inequities
With all that said, we have come to believe that there are two ways our allocation framework should be revised, as we recently described.14 Our perspectives shifted in part from conversations with community stakeholders and representatives of historically marginalised groups. Rather than predominantly grounding civilian triage during a pandemic in the utilitarian goal of accomplishing the greatest good for the greatest number, then adding in equity considerations on the margins, we believe that the proper ethical grounding for triage during the COVID-19 pandemic is similar to the overarching goal of healthcare and public health: to help individuals secure a fair chance to formulate and carry out their conception of a meaningful life over a lifespan.15 Even with a focus on promoting fair opportunity, saving more lives is a valuable goal; it is justified by appealing to the idea that individuals have claims on society to protect their opportunities by avoiding a foreshortened life. All other things being equal, if we cannot meet all of those claims, we should respect as many individuals’ claims as possible.
Accordingly, the first modification we proposed to our triage framework in December 2020 was to correct for the effects of severe structural inequities on individuals’ triage scores using a population-based measure of multiple disadvantage, such as the Area Deprivation Index. We appreciate Schmidt and colleagues’ past and ongoing efforts to encourage policymakers to incorporate this type of correction factor into triage scores. This step is warranted for two reasons. First, it is likely that the two disparity-mitigating criteria we originally proposed—priority to frontline essential workers and the young—would result in ‘patchy’ distribution of disparity mitigation. For example, it is likely that these criteria would result in substantial disparity mitigation among younger individuals and those who are frontline essential workers, but no disparity mitigation among unemployed, older individuals. Second, we believe more potent disparity mitigation efforts are needed. In the time since we originally developed our allocation framework, population-based studies have revealed profound inequities in death rates observed across racial minorities during the pandemic. These inequities arise from unfair distribution of the social determinants of health; there is no evidence that they arise from genetic differences across races. In short, robust disparity-mitigation efforts are needed because the observed disparities arise from unfairness in opportunity across racial groups.
The second modification we proposed is to shorten the time horizon over which near-term prognosis is considered, from ‘death expected within 5 years’ to ‘death expected in less than a year’. There are a variety of validated measures that clinicians should use to aid in accurate prognostication regarding 1-year survival (https://eprognosis.ucsf.edu/). This modification recognises that individuals who have a few years to live have a valid claim on the scarce resources because they stand to benefit considerably from treatment. Moreover, although to date empirical data from the pandemic have not found that a longer time horizon would exacerbate disparities, this is a plausible outcome that prudence dictates should be avoided in light of the deep disparities in COVID-19 outcomes across racial groups. Finally, there is legitimate concern regarding whether physicians can accurately predict outcomes over a 5-year horizon, and whether the imprecision of these predictions may allow bias to leak in against persons with disabilities and other marginalised groups (eg, homeless persons, individuals with mental illness).
Triage decisions during a pandemic are inevitably tragic, particularly because identifiable lives are at stake.16 The most impactful strategies to counteract the consequences of the unfair distribution of the social determinants of health are likely ‘upstream’ from the ICU, such as prioritising disadvantaged communities to receive scarce vaccines and scarce COVID-19 medications, and more broadly by changing social policies to redress structural inequities. However, if triage becomes necessary, triage frameworks should seek to both save lives and equitably distribute that benefit among society’s members.
Patient consent for publication
Contributors DW conceived of and drafted the manuscript. BL provided important intellectual input and provided edits to the manuscript.
Funding This study was funded by Center for Information Technology (Grant number: K24 HL148314).
Competing interests DW reports grants from NIH-NHLBI K24 HL148314, during the conduct of the study; personal fees from American Thoracic Society for work as an Associate Editor of the AJRCCM, personal fees from UpToDate for work as a contributing author, outside the submitted work.
Provenance and peer review Not commissioned; externally peer reviewed.