Changing the equation: Researchers remove race from a calculator for childbirth

Since 2007, obstetricians have counseled patients planning to give birth after a previous C-section with help from a simple calculator designed to determine the likelihood of having a successful vaginal birth after cesarean, or VBAC.

The tool takes into account a patient’s age, height, weight, and their history of vaginal and cesarean delivery. It also asks two yes-or-no questions: “African-American?” “Hispanic?” The answers can predict a drastically lower chance of success for patients of color. But now, after years of work by researchers, advocates, and clinicians, that racialized calculator has been replaced by a newly validated version that is the same in almost every way — except for eliminating race and ethnicity as a risk factor.

The VBAC calculator is just one of several clinical algorithms that have recently been challenged over their use of race adjustment. Providers across specialties have questioned the inclusion of race and ethnicity — which are social, not biological factors — in their decision-making tools, pointing to the risk of perpetuating existing health inequities. But because obstetricians access the VBAC calculator online, it could prove much easier than with other corrected tools to get the updated calculator quickly into use across the country.

advertisement

“I think it’s powerful that this is, in some ways, the first example of race correction being abandoned systematically in a tool in response to these equity concerns,” said Darshali Vyas, a resident at Massachusetts General Hospital.

Vyas first noticed the way race was built into the VBAC calculator in 2017, when she was completing her obstetrics rotation during training at Harvard Medical School. She brought her concerns to a mentor, and in 2019 published a paper with colleagues at Brigham and Women’s Hospital that challenged the use of race in the calculator, which was developed by the Maternal-Fetal Medicine Units (MFMU) Network.

advertisement

“They put it there because the outcomes that are based on societal and structural norms that lead to these differences appear epidemiologically,” said Audra Meadows, a co-author of the paper who was previously the medical director of ambulatory obstetrics at BWH. “But in the end it’s hurting the option to try to have a vaginal birth.”

That, in turn, could disproportionately point patients of color toward a potentially riskier procedure, reinforcing the extreme inequities in maternal medicine: Black women are three to four times more likely to die from pregnancy-related causes than white women, a disparity attributed in part to higher C-section rates. While a repeat C-section is a safer choice than laboring for many patients, a successful VBAC reduces maternal risk for blood transfusion, unplanned hysterectomies, and trips to the ICU.

With the updated calculator, said Meadows, “we may now have another way to promote vaginal birth in a safe way by not convincing someone that they need a C-section based on some calculator.”

It’s unclear whether the original calculator itself contributed to poorer outcomes for patients of color, or that the updated version will change care for the better. The tool doesn’t offer a binary, yes-or-no directive; it only tabulates the odds that a similar patient would have a successful VBAC, a piece of data clinicians can use to inform their conversations with patients.

Still, “it became clear that this was not conceptually something that we wanted to continue having in the model in perpetuity, that it was important not to have a socially constructed variable — race, ethnicity — in a predictive model because of the potential adverse consequences,” said William Grobman, OB-GYN at Northwestern and lead author of the MFMU team that developed both the initial and updated calculators.

“It’s a really critical acknowledgement by the medical community that we got it wrong,” said Michelle Morse, chief medical officer for New York City’s Department of Health and Mental Hygiene and deputy commissioner for the Center for Health Equity and Community Wellness. “And it’s an acknowledgement that we have not always been transparent, honestly, about how the system shapes the clinical questions we’re asking.”

Some obstetricians had already acknowledged those mistakes independently, choosing to forgo entering a patient’s race and ethnicity in the original calculator and defaulting the “African-American” and “Hispanic” answers to “no.”

The analysis behind the updated tool affirms that choice, showing it performs with the same level of accuracy as the previous version. The updated calculator also includes a new, objective clinical variable: whether or not a patient has been treated for chronic hypertension.

“It’s case in point to me that our clinical tools can remain scientifically rigorous and still clinically useful without race correction,” said Vyas.

The updated calculator is far from perfect, Grobman acknowledged. In order to demonstrate that the predictive value of the calculator was the same with or without race, he and his colleagues built it from two-decades-old data on the same set of patient outcomes. And the new calculator has yet to be externally validated in other settings and countries.

Nicholas Rubashkin, a clinical professor of obstetrics, gynecology, and reproductive sciences at UCSF, also noted that the calculator is still a black box. “While this new calculator doesn’t include race/ethnicity, the authors didn’t tell us about the distribution of racial/ethnic groups across the score ranges,” he said in an email. And even a de-racialized tool can reinforce care discrepancies if it is not used carefully. While the calculator was never intended to be a screening tool, Rubashkin’s research indicates some physicians use a VBAC calculator result of 60-70% as a cutoff, refusing to offer the option to patients who score lower.

“There’s still more work that needs to be done to undo the harms of having this tool out in the world since 2007,” said Rubashkin.

Outside of VBAC, work to critically assess the role of race and ethnicity in clinical algorithms is ongoing. In 2020, Vyas and other colleagues highlighted several examples of clinical algorithms that include race correction in the New England Journal of Medicine. In response, House Ways and Means Committee Chair Richard Neal called on professional medical societies to vet their use of race and medicine in their tools.

Since then, medical societies and individual provider systems have made efforts to remove race and ethnicity from their practices. In March of this year, for example, the American Society of Nephrology and the National Kidney Foundation officially recommended ending the practice of race-adjusting the estimated glomerular filtration rate, a measure of kidney function that can play a critical role in whether a patient receives a kidney transplant.

But turning that recommendation into practice will require sustained advocacy on the part of trainees and clinicians, said Vyas. “That eGFR tool is used within labs within individual hospitals,” she said. “There’s a big gap between that recommendation and actually implementing the change.”

The VBAC calculator, though, is typically accessed by physicians online, which could allow for more systematic and faster change. The old tool is still the first hit on an online search for “VBAC calculator.” But its contents have been deleted and replaced with a link to the updated version. While many providers had previously issued internal guidance to omit race and ethnicity from the calculator, enforcing those policies has been difficult; removing the old tool in its entirety does the work for them.

“It’s not our responsibility just to drop something,” said Grobman. “It’s our responsibility to then help disseminate that information.”

For the new tool to maximize its impact, Morse called for more transparency from its developers to acknowledge the motivation for reexamining tools that include race and ethnicity. She also will look for immediate guidance from professional societies like the American College of Obstetricians and Gynecologists to describe how the new tool should be implemented. “I think this has the potential to have more of a systems-level impact,” said Morse. “But the proof is in the pudding.”

And updating the calculator is a small step in the process of building an equitable health care system. Only a sliver of patients have the choice to undergo a VBAC, and many other factors contribute to the disparity in C-section rates and maternal morbidity and mortality. Given the inequities in care entrenched in obstetric practices across the U.S., Meadows is skeptical that changing the calculator will make a meaningful impact on those harms.

“It’s righting the wrongs in medicine more so than improving the outcomes,” she said.

Source: STAT