How racial biases in medical algorithms lead to inequities in care

Read the Full Transcript

• Geoff Bennett:

  Technological innovations are reshaping the U.S. healthcare system, transforming medical science and treatment. But a new report from a coalition of healthcare providers finds that a number of software programs used in hospitals across the country are powered by algorithms with racial biases, which can cause physicians to misdiagnose medical conditions or delay critical treatment.

  Joining us now is Dr. Jayne Morgan, a cardiologist and President-elect of the Southeastern Life Sciences Association. Her work has focused extensively on health equity issues. It's great to have you with us.

• Dr. Jayne Morgan, President-elect, Southeastern Life Sciences Association:

  Thank you. It's great to be here.

• Geoff Bennett:

  And Dr. Morgan, among the biased information these hospitals are using is a kidney function calculation which presumes that black patients have more muscle mass, which can cause delays in care. There's a lung health formula, which uses 19th Century research that suggests that black patients have a lower lung capacity. And then there's a scoring index for determining whether a person can have a vaginal birth after a C-section that gives Black and Hispanic patients a lower success rate. So how do biased data points like these affect patients treatment?

• Dr. Jayne Morgan:

  I mean, you hit the nail right on the head. What you're talking about with a kidney function is something called eGFR, estimated glomerular filtration rate is what we use to do an assessment of your kidney function. And that's incredibly important because that number can determine whether or not you're eligible for transplant, when you should be referred for specialty care, when you may be referred for dialysis which medications you should be on and when you should start them. So can make your kidney function appear better than it really is, and therefore, delay needed and specialty care that you may need to have.

  We see that as well with lung capacity when we look at spirometry. And that in particular is troublesome, because a lot of those calculations are built into the software. And physicians are even unaware that that race calculation is being used in the software, when we look at lung capacity. And that is incredibly important as we are in the middle of this COVID surge as we continue to fight variants. And people certainly can have pulmonary issues, we want to make certain that everybody gets the absolute best care.

• Geoff Bennett:

  Help us understand where this information comes from. As I understand it, it comes from clinical research. But clinical trials have historically contributed to this problem, because they don't tend to have enough people of color involved in them?

• Dr. Jayne Morgan:

  Certainly, there's some historical context to that. And we know that things that were done in the name of research, and I'm saying research in quotation marks from 50, 100, 150 years ago, often began with false premises. And then that was carried on. So, for instance, if we look at the lung capacity, one of the justifications that was used is that body proportions are different in the black race and the white race. And therefore, they put this race factor in that decreases your lung capacity automatically for specific race. And while we recognize that individual proportions may be different, things also can be related to occupational and social determinants of health. And all proportions are not certainly race based. And so, if those calculations are going to be done, they need to be done on an individual basis, and not using race as a proxy in these formulations and in this software.

• Geoff Bennett:

  Dr. Morgan, the bias can extend beyond treatment to medical devices too, is that right?

• Dr. Jayne Morgan:

  So, pulse oximeters are put on your finger to give hospitals and physicians and other health care workers an indication of what the percent of oxygen is in your body, what is your oxygen level. And that's oftentimes can be used to triage you with regard to whether or not you need to be seen immediately, whether you need to be transferred to the ICU, whether you should be given supplemental oxygen, or even intubated. What we know is that the greater the amount of melanin that is in your skin, meaning the color of your skin, it can artificially impair the reading of the pulse oximeter. And so, you can appear if you have darker skin or more melanin to have a higher oxygen level than you actually do. And therefore, you are triaged to lower levels of care when in fact, your medical condition may be more critical than that pulse oximeter indicates. The reason we have that problem is that pulse oximeters were created in the early 70s. And in those clinical trials, they were not inclusive of people of color. We see that as well, with the infrared thermometers that we see, certainly during COVID. People are using them for screening. You place them in front of someone's forehead to get a measure of what is their temperature. And again, we can see that different amounts of melanin in the skin can certainly impair what that infrared thermometer actually reads.

  And so, we've got to begin to make certain that we not only use these clinical tools as physicians, but demand and request that these tools be validated on all peoples of color, because therefore when physicians are using them, they can inadvertently make decisions and also make errors based on readings from software and clinical trials that were not inclusive.

• Geoff Bennett:

  Dr. Jayne Morgan, thanks so much for your time and for your insights.

• Dr. Jayne Morgan:

  Thank you.