In 2020, an NEJM publication highlighted the widespread misuse of race as a proxy for discrimination and genetic ancestry in medical algorithms. That publication and growing social awareness of the harms of racism re-invigorated long-simmering efforts to fix biases caused by the misuse of race.
Since then, researchers and clinicians have worked to update old equations and re-evaluate biased devices in order to reduce health disparities, as independent journalist Cassandra Willyard reported in a recent Scientific American story.
Willyard spoke with AHCJ about how she covered this progress toward health equity.
This interview has been edited for brevity and clarity.
Were you aware of race-corrected equations as an issue before you started reporting this story?
I was aware of some of it. I had been following the eGFR situation and I knew about the pulse oximetry issue. The issues with spirometry were news to me, so that was pretty interesting.
What inspired your decision to cover this topic now?
Even though these problems aren’t solved, the timing of this story was related to a package of stories on advances in health equity. It felt like within this arena of diagnostics, there was progress toward a solution, even if it wasn’t perfect.
What reporting challenges did you experience?
The biggest challenge for me was that I wanted everything to fit into neat little boxes for a story about race correction factors and algorithms. But they didn’t all fit into that structure.
With spirometry and with the kidney function test, the situation is that someone intentionally said, “Okay, there’s these racial differences and we’re going to add this correction factor.” With pulse oximetry, it’s a situation where the technology doesn’t work as well on darker skin colors. The story was focused on diagnostics, so I left out some of the other algorithms based on race, for instance, the equation for VBAC.
It’s a tricky issue to cover, because even with these equations that feel like they’ve been solved, or at least there’s a new race-neutral equation, you can sort of see why the race correction was used in the first place. It wasn’t like people wanted to create disparities; they were trying to alleviate them. The problem was that they picked a bad proxy for genetic ancestry and racism.
How did you decide on the story structure?
I presented them in chronological order, starting with kidney function, and then spirometry sort of followed that. And then the issue with pulse oximetry is one that hasn’t really been solved.
I don’t think these stories could have been intertwined, in part because they were happening in very siloed arenas. There was this larger conversation about race in medicine, but conversations about the kidney function test were happening within nephrology, and about the lung function test amongst pulmonologists. I didn’t feel like I had people who had their fingers in all these different things that I could rely on to make it one single story.
What surprised you?
I have heard a lot about using AI in medicine and I think in my mind, it had always felt like it’s going to exacerbate biases. I think one of the surprises for me was that it can actually help sometimes. I had an example in the story about disparities in treating knee pain, where researchers used AI to examine knee X-ray images and found they could explain the greater frequency of knee pain that Black people experienced. They found something to explain the disparity, whereas before the disparity had just been sort of dismissed. It surprised me to learn that AI doesn’t necessarily have to exacerbate biases.
What stories do you look forward to covering next?
I’m very interested in the AI aspect because I think that is going to become such a big issue in the coming years. It’s going to touch so many facets of medicine, so I’m very curious to follow that piece.
