COVID-19 has imposed on hospitals and health systems around the world the threat of having to ration scarce resources in a way most of us have not experienced in our lifetimes. Demand is outstripping the supply of testing, hospital beds, and ventilators. The dystopian scenario of needing to remove ventilatory support from one patient in favor of another has received the most attention as a source of moral distress. As the pandemic wears on, though, there may be shortages of other resources needed to keep people alive, including hospital staff, blood products, and drugs (such as medications to support blood pressure or palliative treatment for pain and shortness of breath). Demand will also likely exceed the supply of emerging COVID treatments, such as antiviral medications and convalescent plasma. As health systems grapple with how to ensure fair resource allocation and maintain the public’s trust in their decisions, we offer lessons from one state’s experience with health care prioritization.
“Fair” Resource Allocation Can Still Be Biased
In response to concerns about equitable resource allocation, some states have developed guidelines to support fair decision making, specifying objective triage criteria and scoring systems. While these guidelines uniformly proscribe consideration of “social value” in allocating resources, they tend to favor patients in better health and may therefore penalize groups with higher illness burden, including racial/ethnic minorities and people with disabilities, which could exacerbate existing health inequities. Moreover, even the fairest guidelines can only limit the influence of conscious considerations among people applying them, leaving room for unconscious biases to affect decision making.
A large body of research shows that even the most egalitarian among us are influenced by unconscious cognitive processes that make us value certain groups of people over others. An unconscious devaluation of a person’s worth may mask itself as a judgment of worse prognosis. Two clinically equivalent patients may “eyeball” differently due to non-clinical characteristics (such as race, gender, social class, or disability), such that one is deemed to have a better prognosis than the other despite the patients actually having similar prognoses. Embracing the conscious (and morally acceptable) justification of worse prognosis, decision makers may not even be aware that they are engaging in biased decision making. Unconscious bias is particularly likely to influence decision making in stressful situations, when cognitive reserve is diminished and unconscious cognitive processes go unchecked by conscious ones. One can hardly imagine a more stressful situation than having to decide who gets life-saving treatment and who does not.
Even The Fear Of Bias Can Be Harmful
Even if rationing guidelines effectively protected against bias, people from disadvantaged groups might still harbor suspicion. Evidence of health care disparities is long-standing and well-known. African Americans are less likely than whites to opt against intubation and resuscitation in advance directives, partly for fear of receiving inadequate care. In that context, an African American family being told that ventilator support is being removed from their loved one, to support another patient with a better prognosis, might justifiably wonder if the decision was racially biased. These fears are likely to be compounded by emerging data showing wide racial disparities in COVID incidence and mortality, and by reports indicating African Americans are being referred less frequently for COVID testing. Fear of a biased health care system may exacerbate long-standing distrust among disadvantaged groups, potentially deterring them from seeking care and placing them at even higher risk.
Ensuring Unbiased Rationing—Lessons from Oregon
How do we ensure truly fair resource allocation, free of both conscious and unconscious bias? And how do we bolster public trust that decisions will be made fairly? The state of Oregon offers lessons from its extensive experience with health care prioritization. In the early 1990s, the state implemented the Oregon Health Plan, which included a process for explicitly prioritizing which services are provided, and to whom, so that when faced with budget constraints, the state could limit services rather than reduce the number of people covered. This process is managed by the Health Evidence Review Commission (HERC), a group of Oregon health professionals and community representatives. The principles used by the HERC to make priority setting transparent and publicly acceptable in Oregon offer lessons for how to instill trust and protect against bias, and the perception of bias, in rationing care during the COVID pandemic.
First, as recommended by some states, a scarce resource allocation committee should be established for each health care institution or system. These committees should be charged with monitoring resource supply and demand and developing allocation guidelines for resources projected to become scarce. To reduce the influence of subjective judgments in case-by-case decision making, Oregon’s HERC sets resource allocation policy independent of individual cases. COVID resource allocation committees should do the same and should develop guidelines based on clear principles (such as fairness and maximizing benefit) and evidence (regarding predictors of survival, for instance), incorporating objective criteria and avoiding “clinician judgment” as a factor in resource allocation decisions.
Second, when developing resource allocation guidelines, committees should explicitly consider the potential for even objective criteria to exacerbate existing inequities. To protect against this possibility, the HERC explicitly incorporates effects on vulnerable populations in its decision making. Socially disadvantaged groups tend to have more comorbidities and limited access to care, which may result in their being sicker on presentation with COVID. Diverting resources away from those with poorer prognosis may thus have the unintended consequence of further penalizing those who are sicker due to societal inequality. One protection against this type of regressive penalty would be for resource allocation guidelines to use only acute physiologic parameters, rather than chronic comorbidities or functional limitations, in assigning priority scores.
Third, resource allocation guidelines should incorporate public values, ideally with input from diverse constituencies, to ensure that the guidelines themselves are protected against implicit biases. Unconscious bias can affect even third-party decision making, and it is useful to have members whose radar is attuned to spot bias when it creeps in. For this reason, the HERC includes community representatives, including advocates for the elderly and disabled, and receives diverse public input. COVID resource allocation committees should also have diverse representation. Many institutions have Patient and Family Advisory Councils, whose members could serve as a source for community input. Diverse membership may also reduce distrust when difficult decisions are communicated to patients and families who might be suspicious of health care institutions or concerned about bias.
Fourth, resource allocation criteria should be explicit and transparent. Universal awareness of the criteria will increase accountability and reduce the likelihood of decisions being perceived as unfair or arbitrary. Resource allocation guidelines should also be readily accessible and available to patients, families, and communities, to show institutional commitment to fairness and enhance public trust.
Fifth, as recommended by others, application of resource allocation guidelines in individual cases, particularly for the most difficult decisions including diverting ventilatory support, should be carried out by triage officers or teams not involved in the care of individual patients. In addition to relieving clinicians of the emotional burden of bedside rationing, triage teams can be blinded to non-clinical patient characteristics, providing an added layer of protection against bias.
Sixth, resource allocation committee members should be visible and accessible, with publicly available credentials. As social media vitriol has taught us, it is easier to distrust or question the motives of people you do not see or know. Seeing decision makers as real, approachable human beings will reduce the likelihood of resource allocation committees being seen as heartless “death panels.” Accessibility will also allow appeals from clinicians, patients, and families, in cases where considerations not captured in resource allocation guidelines warrant additional review.
Finally, health systems and state and municipal health authorities should stratify COVID tracking data by demographic characteristics and create dashboards to determine, in real time, if disparities in resource allocation are occurring. Any evidence of potential bias should be investigated and remediated. Involving community representatives in monitoring the dashboard would enhance trustworthiness.
We all hope that public health measures will avert the need for difficult resource allocation decisions. But even if we are able to avoid rationing, being prepared with processes to ensure protection against bias will reduce public fear and anxiety. It is also an opportunity to begin to root out and rectify the institutionalized racism and other forms of discrimination that contribute to inequities in health and health care.