Several weeks ago, Americans physicians started receiving reports that our Italian colleagues were being forced to decide which of their patients would receive life-sustaining therapy for COVID-19. Across the United States, American physicians are on the brink of similar devastating choices, if they’re not there already. Some states have started preparing by creating guidelines to help physicians make these choices. There’s already concern that when the time comes, physicians will favor some lives over others, factoring in more than just which patients are the sickest and which ones would be most likely to survive. For instance, will those with intellectual disabilities be denied care? Given how new this viral threat is, it is hard to imagine that we will quickly develop a validated, objective metric to determine who would benefit most from therapy; we don’t have a MELD score for COVID-19. Inevitably, physician judgment will come into play. And that’s scary.

The evidence for implicit biases influencing physician behavior is overwhelming. Implicit biases can be helpful in many respects; they’re part of our instincts as physicians. We wouldn’t be able to function without them. But some biases, such as racial bias, cause harm. We already know that many physicians hold false beliefs when it comes to assessing pain. In one study, half of the white residents surveyed believed “black people’s skin is thicker than white people’s skin” and offered less pain medication to black people. The differential treatment of pain based on race has been confirmed repeatedly over 20 years with no sign of improvements. Another study suggested that physicians with stronger implicit biases favoring white patients had lower rates of ordering thrombolysis for black patients with myocardial infarctions. Black patients are also referred for cardiac catheterization at lower rates than white patients, even lower when the patient is black and female despite presenting with identical symptoms. Even when we attempt to make objective decisions using algorithms, we can fail to account for the bias in the data utilized by the algorithms.

I would not expect that a significant number of physicians practicing in the United States to explicitly state a preference for saving white lives over black lives. I have no doubt that some hold this belief privately, but it’s unlikely they would announce it publicly. Yet the problem of implicit bias in health care goes far beyond those who might hold these pernicious beliefs. Implicit biases affect everyone. All healthcare providers raised in the United States, regardless of their race, are likely to have implicit biases against black people, and that includes black physicians (though the bias is likely not as strong as it is for members of other races). The social milieu of the United States will create these biases in everyone. Our job in health care, in this time of crisis, is to acknowledge these biases and ensure they do not affect our decisions when it comes to choosing who gets the chance to live.

There is already evidence of clustering of COVID-19 cases in parts of New York City where many black and Hispanic people live. Many of these New York City neighborhoods are also economically depressed. If we look back and find that during this period of time, mortality rates among black people are higher than among whites, it will be very easy to suggest this was an effect of neighborhood, poverty, and access to care. Assessing the role bias played in creating these results will be difficult, if not impossible. We will not be able to look back and second guess a physician who chose to intubate one patient over another and determine whether racial bias played a role in that decision. We need to start now by assuming the presence of bias and developing protocols that acknowledge this fallibility in human judgment.

It is incumbent upon us as physicians to be consciously aware of and then account for our racial biases as we treat our patients, not just now, but always. In the immediate circumstance, though, there is a real risk that when asked which patient would benefit most from a ventilator, the same biases that lead us to refer more white men than black women for cardiac catheterization for identical symptoms will unknowingly push the physician to choose the white man for mechanical ventilation. Although the data presented above focus on black and white subjects, we cannot assume that black people will be the only racial group affected by racial biases. With rising anti-Asian racism specifically associated with SARS-CoV-2, Asians and Asian-Americans could also be subjected to biases in receiving care. Racial biases are ubiquitous, and any racial group, especially any racial minority, can be affected.

Implicit biases operate when we are rushing, when we don’t think things through. They are cognitive shortcuts that we all use. The antidote to implicit biases is deliberate thought. I urge health care providers simply to be conscious of how a patient’s race could be influencing their decisions. Ask yourself how (not if) the patient’s race might be affecting you. Acknowledging that racial biases affect us all is our greatest tool for providing equitable care.

Andrew Spector is a neurologist.

Image credit: Shutterstock.com