Introduction

As the novel coronavirus disease (COVID-19) spreads across the USA, many communities are facing the threat of shortages and limitations on healthcare resources, including hospital beds, staff, and critical equipment such as ventilators, hemodialysis machines, and extracorporeal membrane oxygenation (ECMO) machines. In times of crisis such as the current national state of emergency [1, 2], individual states may choose to implement uniform criteria to guide the allocation of scarce resources. These so-called Crisis Standards of Care (CSC) provide a framework for determining who is prioritized to receive scarce resources. At present, the existence of state-level CSC has not been described in the literature, nor have such guidelines been compared at a regional or national level to identify trends or discrepancies that may prove important in the overall health of the nation and that of more vulnerable segments of the population. This study sought to identify and review all publicly available state-level CSC in the USA. Secondarily, we sought to compare the ways in which CSC prioritize patients for the allocation of scarce resources such as ventilators, and to consider the ethical and equity implications of these decisions.

Background

As the H1N1 virus spread in 2009, the Department of Health and Human Services (HHS) Assistant Secretary for Preparedness and Response (ASPR) asked that the Institute of Medicine develop guidelines for resource allocation in times of scarcity. The resulting report [3] led to recommendations and regional workshops [4,5,6] for developing CSC. Key principles on which this framework suggests developing CSC include fairness, duty to care, duty to steward resources, transparency, consistency, proportionality, and accountability [7].

In 2014, the American College of Chest Physicians (CHEST) published consensus recommendations for triage [8] and surge capacity logistics [9] during pandemics and other disasters. Key recommendations include the need for preparation with local and regional stockpiles as well as surge plans across all departments within a health system. With regard to triage [8], CHEST recommends enacting a unified process across an affected geographic area (e.g., state), with appropriate oversight to ensure adherence to the same standards despite potential local differences. They also note that triage processes should rely on protocols, rather than clinical judgment, and that such protocols should be developed in advance of emergencies. Further, they recommend use of an appropriately accurate physiologic scoring system, and reevaluation over time.

As COVID-19 has spread around the world and resources in parts of even the wealthiest nations have become limited, many have again called for explicit planning for the allocation of scarce resources [7, 10, 11]. Within the USA, early data has raised concerns about stark disparities in incidence and case fatality rates among communities of color [12,13,14,15,16]. This has called attention to the ways in which systemic racism may be shaping the pandemic, and has subsequently brought increased scrutiny of existing CSC in states such as Massachusetts [17], where clinicians [18,19,20,21] and politicians [22, 23] alike raised concern about the racial, ethnic, and disability-related inequities that may emerge if and when existing CSC are implemented. Massachusetts’ CSC have since been updated [24] in response to this advocacy, incorporating more specific language around health equity but retaining the same decision-making structure.

This review was undertaken in an effort to better understand the current status of CSC around the country, including the ways in which existing guidelines instruct physicians and healthcare systems to allocate specific resources such as ventilators, which may become scarce during a widespread respiratory illness such as the present COVID-19 pandemic.

Methods

Two authors (ECM and CS) attempted to identify CSC for each state between April 13 and April 17, 2020. Identification of state-level CSC was first attempted using a list available on the ASPR Technical Resources, Assistance Center, and Information Exchange (TRACIE) website [25]. If no document or only a nonspecific CSC was found, Google was used to search using the terms “state name” and “crisis standards of care.” If the first ten search results failed to yield a specific CSC, the state’s department of health (DOH) website was searched, with particular attention to Preparedness, Emergency Response, and COVID-19 pages. If this, too, failed to identify a guideline, the DOH was contacted via the website’s information request form, and/or via a personalized email from ECM or CS to an individual from the DOH’s Preparedness (or equivalent) team. If an email address for a relevant DOH employee was not easily identified but a telephone number was listed, the DOH was called to request information about CSC for the state.

Available CSC were read by the first two authors and categorized as “specific” if they include explicit decision-making criteria for allocation of resources including ventilators, or “nonspecific” if they lacked clear decision-making criteria. Reviewer agreement on this distinction was 100%. CSC were systematically examined for the following content: explicit ethical principles around which guidelines were created; whether equity was explicitly listed as a guiding principle (e.g., explicit discussion of equity between individuals of different identities, marginalized groups, or discussion of population-level disparities); whether CSC explicitly state that resources should be allocated in an identity-blind manner (i.e., without regard to race, ethnicity, gender); specific guidance for allocation of critical care resources, including ventilators; exclusion criteria for access to critical care; criteria used for prioritization of patients for critical care resources, i.e., use of SOFA [26] or Modified SOFA (MSOFA) [27, 28] scores; and consideration of long-term comorbidities in the prioritization framework. Reviewer agreement on each of these factors was greater than 90%; discrepancies were adjudicated by discussion. Descriptive statistics were used to summarize characteristics of identified CSC.

Results

A state-level CSC document was identified for 29 states (see Fig. 1 and Table 1). Fifteen were identified from the ASPR TRACIE website [25]. Six of the 28 state-level resources listed at that site (CA, FL, MD, MT, OH, TN) were either not available via the provided links, or were deemed to not be CSC as they did not include guidance for triage and/or resource allocation. Updated versions of several other CSC documents in that list were identified by the authors (e.g., MA and PA). An additional ten CSC were found through electronic searches either using the search terms noted above or on the state DOH website. A total of 22 state DOH were contacted by email, 14 of whom replied on or before May 3, 2020. Four additional CSC were identified after email correspondence with state DOH personnel, and six states confirmed that CSC were in development. For the three states without an identified email address, as well as two for which only a generic inquiry email address was available, state DOH were contacted by telephone; these calls did not yield any additional CSC documents (Fig. 2). Appendix 1 contains links to each of the reviewed CSC documents.

Fig. 1
figure 1

Crisis Standards of Care guideline identification process

Table 1 Comparison of available state-level Crisis Standards of Care (CSC)
Fig. 2
figure 2

Crisis Standards of Care across the USA, by status of development as of May 3, 2020

Wide variation was noted in the degree of detail provided in CSC regarding the process of their development and the principles on which they were created (Table 1). Fifteen CSC (51.7%) were created or updated this year; eight (27.5%) specifically addressed the COVID-19 pandemic. Twenty-four (85.7%) explicitly stated the ethical principles on which resource allocation decisions should be made. Nineteen (65.5%) explicitly articulated that resource allocation decisions should be made without regard to race, ethnicity, disability, and other identity-based factors. Health equity with respect to these identities, not including equity among hospitals, was identified as an ethical consideration in 16 of the available CSC (55.2%).

The means by which scarce resources should be allocated under each of the 29 identified CSC varied widely (Table 2). Three states (KY, MI, WY) explicitly left all triage and decision-making to local institutions. Ten CSC (34.5%) incorporated exclusion criteria which would preclude patients from access to critical care when CSC are enacted (see Tables 2 and 3). The most common exclusion criteria (Table 3) included low likelihood of immediate survival, e.g., ongoing cardiac arrest (80% of states with exclusion criteria), neurologic conditions including “severe dementia,” or “advanced” and/or “irreversible” neurologic events (70% of states with exclusion criteria), and “severe” or “overwhelming” trauma or burns with very poor (e.g., < 10%) predicted survival (80% of CSC with exclusion criteria). The degree of specificity provided within each category varied widely, e.g., “CHF (NYHA Class III or IV), left ventricular dysfunction, hypotension, new ischemia; known congestive heart failure with ejection fraction less than 25% or persistent ischemia/pulmonary edema unresponsive to therapy,” in one CSC (AL), compared with “severe congestive heart failure” as a criterion for transfer to palliative care rather than ICU admission during periods of resource limitation (WA).

Table 2 Comparison of available state-level Crisis Standards of Care guidance for the allocation of critical care resources
Table 3 Comparison of exclusion criteria for access to critical care from available state-level Crisis Standards of Care

Of the 29 identified CSC, ten (34.5%) allowed for the consideration of societal value in prioritizing health care workers; four used essential worker status as a tie-breaker for otherwise equally prioritized patients, while the remaining six did not specify exactly how essential workers might be given priority (Table 2). Five (17.9%) CSC factored pregnancy into decision making, either as a general consideration (OR), explicitly in algorithms by increasing priority for pregnant patients, particularly if at a gestational age compatible with fetal viability (MA, PA, UT), or as a tie-breaker if other factors are equal (CO) (Table 2). Specific processes for ventilator allocation were discussed in 21 CSC (72.4%). All 21 states with specific frameworks for allocation of critical care resources recommended SOFA [26] or MSOFA [27, 29] scores as a component of prioritizing patients for allocation of resources. Among the 21 specific CSC frameworks, 15 (71.4%) incorporated consideration of pre-existing chronic health conditions/comorbidities which affect long-term survival (Tables 2 and 4). Within the 15 CSC that recommended consideration of comorbid conditions in prioritizing patients for resource allocation, the most frequently cited comorbidities included heart failure or severe coronary artery disease (53.3%), chronic lung disease (60%), end-stage renal disease (46.7%), cirrhosis or other end-stage liver disease (53.3%), neurologic disorders (e.g., Alzheimer’s dementia) (33.3%), and active malignancy with poor prognosis (46.7%). The specificity with which each of these comorbid conditions was described varied between states. For example, Utah listed “end-stage COPD” while other states included “home oxygen dependent” (AK, MN, VT) as descriptors of the degree of chronic lung disease that should be factored into prioritization considerations. The Modified Charlson Comorbidity Index [30,31,32] was used by Colorado as a means for accounting for comorbidities in prioritization of resource allocation.

Table 4 Comparison of long-term comorbidities included in prioritization framework from available state-level Crisis Standards of Care

Discussion

The allocation of scarce healthcare resources during times of crisis is fraught with ethical challenges. Development of CSC has the potential to mitigate both individual and societal moral injury by establishing a clear framework for decision-making and prioritization. This is of particularly urgent concern due not only to the rapid progression of the COVID-19 pandemic in the USA and impending resource shortages, but also due to increasingly widespread concerns about racial inequities that are emerging from states and communities that are tracking demographic data for patients with COVID-19 [12,13,14,15,16]. This review endeavors to identify and compare all publicly available, state-level CSC as SARS-CoV-2 continues to spread throughout the USA.

The development of CSC was identified as a priority by the federal government, which has provided strategic guidance for how to do so [3,4,5,6, 11]. Although some have suggested that as many as 36 states have developed CSC [33], we identified only 29 states with publicly available CSC as of May 3, 2020. The degree of specificity and detail in available CSC is extremely varied, ranging from broad ethical guidance to shape the development of local- or hospital-level policy, to explicit criteria and algorithms for determining the allocation of ventilators and other specific resources. The general structure of these algorithms includes consideration of the following: (1) exclusion criteria, usually based on a low likelihood of survival despite maximal resource allocation; (2) a calculation of an objective score to reflect the severity of present illness and thus prioritization category; and (3) repeated evaluation over time to determine ongoing priority status. In some cases, additional consideration is made for factors that may predict individual or societal benefit. Most algorithms explicitly state that the physician responsible for direct patient care should not be the same individual making decisions regarding allocation of scarce resources to that patient. It is essential that the equity implications of CSC content at each level—exclusion criteria, use of SOFA/MSOFA as a marker of acute illness severity, and the consideration of comorbid conditions and other individual factors such as social worth—be critically examined before CSC are implemented. Similarly, the process by which CSC are developed must include diverse perspectives to ensure resulting documents are not inadvertently discriminatory.

Crisis Standards of Care: Content

Exclusion criteria are intended to avoid allocating scarce resources to patients who are unlikely to survive regardless of the care provided; however, when CSC include broad, nonspecific exclusion criteria, their application may introduce significant bias or variability at either the clinician or institutional levels. For example, the Kansas CSC [34] had broad exclusion criteria including renal failure requiring hemodialysis, and metastatic malignancy with poor prognosis. In contrast, New York’s CSC for ventilator allocation [35] explicitly did not exclude patients based on the need for dialysis or the presence of metastatic malignancy, a decision which their guidelines report was made due to wide variations in prognosis, life expectancy, and quality of life among individuals in these groups. If exclusion criteria for accessing critical care are used, it may be that very specific guidance are more useful than broad recommendations for excluding those with “poor prognosis,” as interpretation at the facility or individual level may lead to inequitable exclusion of some individuals.

All 21 CSC that specifically discussed critical care and ventilator allocation relied on the SOFA/MSOFA score as an objective means of evaluating the severity of illness at the time of triage although SOFA scores were developed for use in the study of populations, and are thus not accurate for predicting individual-level mortality [36]. Some studies have raised concerns with the validity of these scoring systems as a triage tool, given that more than half of patients with the highest category of SOFA or MSOFA scores (> 11) ultimately survived [27]. Nevertheless, most CSC used SOFA scores to categorize patients into one of four priority levels. Most CSC explicitly counsel against the comparison of two individuals within the same category based on a slightly differences in the exact SOFA score, although some suggest using the numeric score as a tie-breaker. Further research is needed to better understand the ways in which this could disproportionately affect different communities.

A great deal of variation emerged when comparing the extent to which CSC incorporated consideration of factors beyond the severity of acute illness requiring ICU care. Of the 29 available specific CSC, 19 (5.5%) explicitly stated that triage decisions should be made in an identity-blind manner (i.e., without consideration of race or ethnicity). Eleven of these CSC, however, are among the 15 states (see Tables 2 and 4) that considered comorbid conditions in resource allocation decisions. The consideration of life-limiting comorbidities in allocation of resources during times of crisis may contribute to maximizing the number of lives, and/or the number of life-years saved. However, given the well-described and profoundly unequal distribution of illness between racial and ethnic groups in the US [37,38,39,40,41], the consideration of comorbid conditions such as hypertension [42], diabetes [43, 44],, and chronic kidney disease [45], is likely to function as a proxy for allocation of resources by race and ethnicity. Similarly, individuals with disabilities have a greater incidence of comorbid health conditions [46], and have more limited life expectancies [47] than able-bodied individuals. Consideration of long-term prognosis in allocation decisions may prove problematic, as CSC from Pennsylvania acknowledge: “Based on consultation with experts… we have intentionally not included a list of example conditions associated with life expectancy <1 year and <5 years… [to avoid decisions] being applied as blanket judgments, rather than in the context of individualized assessments by clinicians, based on the best available objective medical evidence.” [48]

Given the significant disparities in incidence, morbidity, and mortality from COVID-19 that have emerged among minority populations [12,13,14,15,16], the consideration of identity-based characteristics and their proxies must be carefully weighed in the development and implementation of CSC. Clarity is needed as to why specific comorbidities should lead to exclusion or lower prioritization for resources. There may be a fundamental ethical difference between excluding a patient because their condition significantly reduces long-term life expectancy, and excluding a patient because their condition significantly reduces their likelihood of surviving the acute illness that has led to their need for life support. But that distinction alone may not prove dispositive. How likely their survival is in the short term may matter as well. Giving equal priority to certain conditions that reduce short-term survival to some degree but are highly linked with race or socio-economic status, for example, may be justifiable on equity grounds. Yet as short-term survival becomes increasingly poor for certain comorbidities, arguments for equal access to scarce equipment may prove less convincing. Consensus on these challenging questions remains elusive. However, CSC that elucidate the rationale for these choices enable more meaningful equity analysis and are more conducive to widespread acceptance. In the absence of compelling reasons to include consideration of specific comorbid conditions, particularly if these are inequitably distributed between racial and ethnic groups as a result of persistent structural racism within American society, CSC would more equitably distribute resources by using a lottery system to break ties between patients who are equally acutely ill. Although none of the reviewed CSC suggested using a weighted lottery, the process recently developed for the allocation of scarce medications (e.g., novel antivirals) at the University of Pittsburg provides an interesting strategy for prioritizing more socially vulnerable individuals [49]. Further research is needed to determine the equity implications of such a strategy.

Crisis Standards of Care: Process

The development of CSC, while critically important to support clinicians and healthcare systems during times of crisis, must not be undertaken in a rushed or siloed manner. Failure to incorporate broad perspectives and public comment may lead to oversights or unintentional consequences. These may ultimately lead to CSC needing to be revised. For example, due to the equity concerns raised by clinicians, politicians, and disability activists, the Massachusetts CSC were revised 2 weeks after they were first published [17, 23]. Input from diverse individuals and stakeholders is critical to ensuring that CSC reflect the values of the populations they will affect. While ethical frameworks for allocation of scarce resources seek to provide the greatest good for the greatest number, interestingly, this may be at odds with what the general public may classify as a more equitable way of distributing in times of a pandemic [50]. Arriving at consensus with input from diverse stakeholders, with particular attention to the inclusion of historically marginalized groups, may prove crucial to ensuring the acceptance of CSC and the consequences of their implementation in the long term.

Failing to provide any guidance at the regional or state level may prove even more problematic than developing CSC with insufficient community input. Without objective and concrete guidance for resource allocation during times of scarcity, “first-come, first-served” remains the default triage process, thereby allowing patients who arrive sooner to receive care, even if they have a far lower likelihood of survival than someone who arrives shortly thereafter. Most people, including ethicists, agree that such a system is not in society’s best interest [51, 52]. Not only does this method fail to maximize the number of lives saved or allocate care in a health-preserving manner, but it raises grave equity issues of its own. Socio-economic factors may increase healthcare literacy and thus awareness of the need to seek care earlier during a crisis, while lower-income workers may not be able to leave their jobs to line up first. In addition, well-resourced patients may travel farther to seek hospitals where lines for critical equipment are shorter.

The absence of state-level guidance places the burden of decision making upon hospitals and often upon individual providers. Fewer than half of the hospitals in the country appear to have developed CSC [53]. At present, the majority of hospital administrators and physicians in the USA are upper-middle class (or wealthier) and white [54]. No matter how good their intentions, they are likely to be blind to biases built into protocols designed without external input. Clinicians in the field, forced to make rapid decisions, will inevitably bring their own biases—conscious and unconscious—into the triage process. This may affect both patient assessment and resource allocation decisions. These biases, if unchecked, may result in stark inequities in both care and survival at the population level.

In addition, leaving doctors in the field to make decisions about triage criteria is bound to take a significant emotional and psychological toll. This toll may weigh more heavily on physicians of color and those from lower socio-economic backgrounds if they consistently witness patients with identities similar to their own losing out in the triage process. Finally, it is worth noting that optics matter: Even if the allocation were to prove equitable, the long history of systemic and structural racism in the USA may lead patients and families who do not receive priority to conclude that bias by the individual hospital or physician was a factor in their assessment, a concern which can be mitigated—at least to some degree—by a region or state-wide policy.

Conclusion

There is wide variability in the existence and specificity of CSC across the USA. Although many states have recently updated or developed CSC, and others are actively working to do so, these processes should be collaborative with broad input from the affected community. Many of these decision-making criteria have implications for health equity. CSC may disproportionately impact historically disadvantaged populations due to inequities in comorbid condition prevalence, expected lifespan, and other systemic disadvantages that result from structural racism within American society. Failure to consider these factors will inevitably lead to the perpetuation of structural and historical inequities through the application of these standards.