Race-Based Prediction in Pregnancy Algorithm Is Damaging to Maternal Health

Race-Based Prediction in Pregnancy Algorithm Is Damaging to Maternal Health

Algorithms Abound

In today’s fast-paced health care system, algorithms are ubiquitous, and the provision of maternity care is no different. Maternity care algorithms intend to systematically assist providers in the clinical management of a variety of problems that can occur during pregnancy. But these algorithms are used alongside a backdrop of vast racial disparities in maternal mortality and morbidity in the United States; Black women nationally are three to four times more likely to die from pregnancy-related causes than white women. Clinical tools, including algorithms, may exacerbate these disparities if not properly scrutinized. One such tool is the Vaginal Birth After Cesarean (VBAC) calculator, which estimates success rates for vaginal birth among pregnant people with a previous cesarean delivery.

VBAC Is an Individualized Decision

For pregnant people with a prior cesarean delivery, some safety concerns exist with a subsequent vaginal delivery, especially for people with more than two prior cesarean deliveries, prior uterine surgery, or prior classical incisions. However, there are also well-established health benefits for a successful VBAC compared to a repeat cesarean delivery, including avoidance of surgery, lower risk of complications, and faster recovery time. As a result, clinicians use tools like the VBAC calculator, endorsed by the National Institute of Child Health and Human Development, to help providers determine how risky it would be to try for a VBAC.

The VBAC Calculator

Until May of this year, the VBAC calculator had two race-based correction factors that “subtracted” from the overall likelihood of successful VBAC, meaning pregnant people identified as African American or Hispanic were systematically assigned a lower chance of successful VBAC than all other women. We know that providers are influenced by concerns over liability and perceived risk when counseling patients about labor after cesarean section. So because the calculator dissuaded clinicians from offering labor to pregnant people with low VBAC scores, the algorithm likely exacerbated racial disparities. In fact, research shows that the VBAC calculator underestimated the actual success rates for Latinas.

Inclusion of race in the algorithm was based on observational data, which represented a snapshot in time. And it’s not surprising that a snapshot would reveal racial and ethnic disparities, which are widespread in maternal health. But incorporating race-based correction into the calculator systematized these existing disparities into a predictive algorithm, thus ensuring that the trends would continue and perpetuating institutional racism.

Removing Race from the Calculator

Two months ago, investigators published research in the American Journal of Obstetrics & Gynecology validating a VBAC algorithm that did not include race or ethnicity. This resulted in a long-overdue change to the online calculator available through the Maternal Fetal Medicine Units Network. The new version removes questions about race and ethnicity, and adds one about the patient’s history of hypertension.

The example of the VBAC calculator presents a prime opportunity to more closely examine race in algorithmic decision-making. While it is critical to acknowledge racial and ethnic inequities in maternal health outcomes, the inclusion of race as a factor in predictive algorithms has the potential to worsen existing disparities. The VBAC calculator is just one example of racism in reproductive and sexual health decision-making, and other algorithmic tools that include race-based correction factors, such as the Breast Cancer Surveillance Consortium Risk Calculator and the National Cancer Institute Breast Cancer Risk Assessment Tool, need to be similarly reconsidered.


Additional Resources

Race-Based Prediction in Pregnancy Algorithm Is Damaging to Maternal Health

Preventing Harm from Automated Decision-Making Systems in MedicaidCommon

NHeLP AHRQ Comments

Demanding Ascertainable Standards: Medicaid as a Case Study

Q&A: Using Assessment Tools to Decide Medicaid Coverage

Ensuring that Assessment Tools are Available to Enrollees

Medicaid Assessments for Long-Term Supports & Services (LTSS)

Evaluating Functional Assessments for Older Adults

Opportunities for Public Comment on HCBS Assessment Tools – National Health Law Program

A Promise Unfulfilled: Automated Medicaid Eligibility Decisions

Cases

A.M.C v. Smith

Darjee v. Betlach

Hawkins v. Cohen

LS v. Delia

Related Content