An interview with Yekki Song
As part of Vital’s series on Racism in Science, we interviewed medical humanities researcher Yekki Song, a 6th year MD-PhD student at the University of Texas Medical Branch in Galveston, where she studies how society shapes and influences the practice of medicine.
We spoke with her about a newly-published paper she co-authored in the Journal of Pediatric Surgery, one that caused her to begin questioning the way in which she and her colleagues had used race as a key predictor of health outcomes
Before you tell us about the study you worked on, can you tell us a little bit about who you are, and how you got into this line of research?
I’m currently a 6th year MD-PhD student in the medical humanities. I started off in medical school studying pharmacology and toxicology, but I was always really interested in health disparities, particularly racial health disparities, and wound up switching programs.
This particular project began when I was working with a pediatric surgeon to mine data in a massive medical database, one with about 21 million entries, to look at health disparities in post-surgery outcomes for babies with gastroschisis. That’s a condition where the abdominal wall of the baby doesn’t form properly, so its viscera kind of “hang out” in the amniotic fluid.
Why gastroschisis? Did it seem like a likely candidate for further inquiry?
It was a likely candidate because it’s one of the most common congenital defects, so we weren’t worried about an underpowered study. Plus, since babies are born with this condition, which develops in utero, they go straight from birth to NICU for gastroschisis repair, so we thought that by controlling for more at-home factors we’d be able to attribute more of the differences to things that happened in the hospital.
One aspect of your study was prenatal care, so were you looking at when the condition begins to develop? What exactly were you trying to find?
It can be diagnosed in utero through ultrasound—there are specific ultrasound markers. And there are many studies that look at very specific early ultrasound markers and whether they have any correlation with how severe the condition winds up being.
Our study, though, was looking at very specific health outcomes for babies after their repair surgeries had been performed. To do that, we looked at diagnoses codes associated with their hospital stay, trying to find patterns between various risk factors and the complications that did or didn’t develop in various patients.
There are risk factors for having the disease, and also for getting a more complex version of the disease. In the simple cases, it’s a pretty basic surgical fix, and the baby is fine. But with the complex version, there are complications, they have trouble with feeding, and stay in the hospital for a longer time. So we were looking at the difference between simple and complex gastroschisis, and one thing we found was a disparity in the complications experienced between infants of black mothers versus those of other races.
We did regression studies that controlled for different hospitals, socioeconomic factors, etc. People who identified as black were associated with increased odds of intestinal atresia, a feature of complex gastroschisis, and also associated with increased morbidity. So, we found it alarming that there was an increase in incidence and an increase in worse outcomes.
But then, at a certain point in the process you became uneasy about how you and your colleagues were using race as a determining risk factor. So, what was your basis for looking at race in the first place?
Just to clarify, incidence of gastroschisis is actually higher in white mothers, but one study had found that there’s a disproportionate increase in children of black mothers age 20 and under. That was part of the justification for looking at race as a risk factor: why did black mothers have more gastroschisis than they ever did before?
However, we weren’t actually aware of that study until a few months into our study. It was a post hoc justification for looking at it. We originally looked at race only because we were expecting to find a racial disparity—which exists almost across the board—with gastroschisis.
Okay, but where’s the problem in that, exactly? Is the problem that your study didn’t take into account other risk factors, like various forms of racial bias in health care?
Well, there was that. We did try to address that in our paper. We talked about needing to look for provider-level bias, and making sure that babies of non-white mothers aren’t put at risk for complex gastroschisis due to implicit bias. But the problem that began to dawn on me was actually a lot more complicated than that.
These babies were also more likely to be pre-term, which is also associated with increased atresia. You might think that this is because the mothers were young, but there have been some studies suggesting that black women are more likely to give birth prematurely regardless of socioeconomic status, because of something called allostatic load, which is the accumulated effect of chronic stressors that influence a mother’s DNA and get passed down. This likely has to do with institutional racism, but the thing that becomes so tricky is that it’s hard to prove this in the scientific literature. You can’t do a randomized study proving that racism is related to allostatic load.
It sounds like one of the problems you face as a researcher is designing studies that account for some of the systemic reasons behind why people who identify as black might have certain risk factors without reinforcing already racialized assumptions that there’s something bad about being black.
I think that’s a fair description. Yes, there is a balancing act. I was trying to be conscious of not reifying race as a biological category. But you have to look at race as something that has influence. You have to be very careful when you present your research. This is an association and not a causal relation. It is a really tricky delineation. When you set up your logistical regression, and you set up race as a prediction value, it’s important not to say that simply being black is a risk factor. You have to be very clear.
The subtlety of this distinction doesn’t always make it into the headlines when such studies are reported.
Yeah, most scientists aren’t very good at media coverage, so the media can get away with sensationalizing findings. Also, doctors tend to take studies about risk factors, and instead of thinking about it from a population level, they individualize it. That’s a problem, because large scale population studies do not translate neatly into individual risk factors.
Do you worry that there’s a kind of feedback loop problem between researchers and doctors? Doctors report patient data in a particular way, which you mine for patterns and report on in ways that are inevitably structured by the form of data gathering, and those findings then inform doctors who try to apply it in clinical practice, and then report their data to scientists . . . and so on. Does that make sense?
I don’t know about a feedback loop, but I do want to figure out the ways in which scientific research falls short. You mentioned the data sets I had to work with, and one thing that I couldn’t figure out was how race had been determined in the database. There were only four categories given: white, black, hispanic, and other. Already those categories are very limited and problematic, since “hispanic” is an ethnic category, not a racial one. And there was no way of knowing who decided what “race” the patients were. Were they asked to self-identify, or did a doctor just visually assess and check the box?
If we were to move out from talking about this particular study, what is it like more broadly to be a scientist interested in race and racism?
It’s very difficult to talk about race from a scientific perspective. It is hard to “prove” that different races of people are treated differently. You can have researchers who do not understand or believe in institutionalized racism. You have to make it past co-authors and reviewers—if you can’t prove it, then it isn’t included. So, there are limits from the scientific side.
As medical humanists, though, we can really articulate our work: calling for better research on race that doesn’t reify race as a biological category. Really getting into conversations with scientists about why the research is flawed, and what underlying assumptions are being made.
What’s the best angle of approach for medical humanists to try to make those changes? I’m guessing that there’s limited effectiveness in couching the questions you want to pose in political terms.
Adopting the language of scientists would help. There’s so much good literature out there on race, but scientists aren’t reading it. Many don’t know it exists. We need to bring the literature that we have to scientists. That could take many forms. It could be just as simple as sending an email to the authors of a study, offering them some recommended reading.
Speaking of language, I noticed you mentioned biocitizenship when first talking with us about your research. That’s a term from social science. Is it a concept that’s well known among medical researchers?
No, it’s not very well known in science literature at all. Biocitizenship is the imperative to be a moral citizen by participating in certain biological acts. So, becoming a mother at a young age could be considered by some a breach of good biocitizenship. I found it remarkable that researchers often lay out assumptions about under-twenty mothers: that they are irresponsible, unwed, lacking prenatal care, etc. None of these potential factors are necessarily true.
It seems like this might be an especially difficult time to be trying to put these issue on the radar of scientists, since so many in the scientific are already feeling under assault from the right. Is it a challenge for more progressively-minded parties to ask for their attention as well?
It is a difficult time. The key is not to be critical, but to critique. It’s not easy to find ways to challenge racial bias, obviously, but it is important.
This interview has been edited for length and clarity.