Health Care Anti-Bias Tools May Be Endangering Classes of Essential Workers

Part of the Series

Despair and Disparity: The Uneven Burdens of COVID-19

Bias is so baked into health care delivery that COVID-19 forces us to rethink how we allocate resources so Black and Brown people don’t suffer more than they already have.

Health care providers worldwide are now considering how to most fairly allocate scarce resources in the wake of the coronavirus pandemic. The thought of having to use an algorithm to determine who gets access to life-saving resources and who does not is unquestionably difficult to imagine. We must support the efforts of clinicians, scientists, and engineers who are developing and deploying more and better equipment so that greater numbers of patients can be treated, avoiding the devaluing of any life based on race, age, ability status, etc. Until we have developed a medical infrastructure that treats all individuals fairly, algorithms may be the most objective way to allocate resources. Therefore, we should examine these algorithms and think critically about the fairest and most just way to allocate resources should the unthinkable arise.

Most rationing guidelines agree on a utilitarian principle of maximizing benefit for the most people. These criteria prioritize health care workers and use purportedly objective metrics like the Sequential Organ Failure Assessment (SOFA) score, which is based on the dysfunction of multiple organ systems, and could be used to allocate scarce resources to patients who have better organ function and thus a greater likelihood of good outcomes.

Even if hospitals aren’t prioritizing health care workers or using mortality predictors like the SOFA score, most algorithms are based on the principle that those who are more likely to have a poor outcome should not receive limited resources. Some approaches have called for allocation by lottery, throwing out utility in the name of equity, but even these approaches can disadvantage people of color if there are pre-existing inequalities in health care access and utilization across races. Clearly many of these metrics have hidden, implicit biases against people of color, as disproportionately poor outcomes for African Americans, Latinx and Indigenous populations bear this out. We’re kidding ourselves if we don’t rethink what “fair” means across all patient populations.

People of color are particularly disadvantaged by resource allocation protocols. One recent study found “black people who visited hospitals with COVID-19 symptoms in February and March were less likely to get tested or treated than white patients.” In some communities, especially in Western states, Latinos account for a large proportion of COVID-19 cases despite being a minority in population size. And some Native American tribes have higher infection rates than most states, even including hard-hit areas like New York.

While resource allocation policies largely have not been activated in the U.S., it is likely they will be needed in the future. We may face a new wave of coronavirus infections or even a different pandemic. These decisions are made using algorithms to determine which patients receive access to life-sustaining medical resources should the immediate need for them outstrip supply. States and hospital systems are creating or updating these policies based on recommendations from ethicists and medical organizations. Two principles frequently employed in resource allocation policies include rewarding “instrumental value,” such as giving preference to health care workers, and the likelihood of surviving without having major or long-term deficits via SOFA score.

Sadly, these criteria may also foster implicit bias.

For example, why prioritize health care workers? Once recovered, they can continue to help save the lives of others, the reasoning goes. There is also an element of “responsible stewardship,” where asking health care workers to risk their lives to save others only to be denied treatment if they fall ill, is unfair. This is logical on its face. But applying this principle has two major flaws.

First, “health care workers” is an ambiguous term. Is the definition limited to clinicians directly providing medical care, such as physicians and nurses? Or might it be extended to other staff in medical facilities, like food service workers? Maybe essential workers — such as grocery store employees and supply chain truck drivers who cannot work from home — should be included. After all, they are exposed to the coronavirus at higher rates than the general public.

Here’s where more implicit bias comes in: Black and Latinx people are underrepresented as physicians and registered nurses. At the same time, African Americans make up the largest percentage of health care support staff. If treatment preference is given to clinicians only, a large number of hospital staff who provide critical services will not get priority access to treatment. This is not only unethical, but it is also in direct opposition to the principle of maximizing benefit. Inappropriate application of the label “health care worker” could lead to putting the needs of a largely white group first, while ignoring Black health care workers on a different hierarchical rung.

Black and Latinx people are less likely to be able to work from home than whites and Asians. Black and Brown workers also make up large numbers of workers in food preparation and service, and material production and transportation industries. People of color, for example, are concentrated in lower-paying jobs in the restaurant industry. These occupations have proven over and over again how essential they are. If resource allocation policies prioritize health care workers but not other essential workers, Black and Brown people will continue to be exposed to higher risks, without receiving benefits of responsible stewardship.

Maximizing benefit is at the foundation of putting first those patients with the greatest likelihood of having positive outcomes. To many, it makes ethical sense to provide resources to those who have the greatest likelihood of recovering and those who, once recovered, would have the longest life expectancies. Because these metrics are impossible to predict, many protocols turn to validated scoring systems, such as SOFA scores, as a less-biased way to assess. One study found that a full 80.8 percent of hospitals surveyed were using these scores or a similar scale.

However, this approach has been criticized as being unfair to elderly patients, who have lived longer, and thus, have fewer years to live, as well as patients with disabilities, who have less access to preventative care and subsequently poorer health outcomes. These patients tend to have more comorbidities, and thus, higher scores.

Black and Brown people also are disadvantaged by this scoring system.

While the gap is closing, African Americans still have the lowest life expectancies of any race in the United States. The SOFA score, meant to quantify organ dysfunction across six systems (respiratory, cardiovascular, hepatic, coagulation, renal and neurological), puts Black and Brown people at a disadvantage. Both Black and Latinx populations have higher rates of cardiovascular disease, hypertension and diabetes. As a consequence, something like the SOFA score, which attempts to predict morbidity and mortality based on the failure of these organ systems, may perpetuate or even magnify the health outcomes disparities experienced by people of color, who already have higher morbidity and mortality rates. Failure to factor in these facts of life is how implicit bias seeps into attempts at just decision making.

When adding up the ways we prioritize who gets help first, the equation doesn’t add up for historically underserved populations. Some health care professionals have pushed back against these resource allocation policies, arguing for justice-based algorithms that take into account demographic and social data in order to ration resources in a more equitable manner. Revamping these policies to make them fair for all populations demands our serious time and effort.

Unfortunately, without algorithms, judgment is left entirely up to bedside clinicians, and patients receiving scarce, needed assistance are often chosen based on the personal, inherent biases of the clinicians involved. Beyond resource allocation policies, COVID-19 has exposed the systemic inequality that has pervaded health care and has disadvantaged people of color for far too long. The current pandemic should serve as a wake-up call that it is time to reexamine how we provide care at its most fundamental levels so that health care inequity is not perpetuated or tolerated any longer. Before the next surge occurs, we must develop better standards for sharing resources to root out deadly bias.