Report: Rethink Race-Based Adjustments in Clinical Tools

New guidelines are out for researchers and others for how to use information about race and ethnicity in biomedical research, with implications for practicing clinicians.
Race correction or race adjustment — the practice of developing clinical calculators or guidelines that modify their output based on race or ethnicity — can have profoundly negative effects on patient care, according to a 2024 report from the National Academies of Sciences, Engineering, and Medicine.
And despite more recent corrections to various clinical algorithms and decision support tools, uptake of changes is slow, according to the report and other experts.
The report cited the risk model for ovarian cancer, the colon cancer survival calculator, the atherosclerotic cardiovascular disease risk calculator, and the fracture risk assessment tool, among others, as incorrectly incorporating race and ethnicity, influencing diagnosis and treatment decisions.
Sindy Escobar Alvarez, PhD, program director for Medical Research at the Doris Duke Foundation, which commissioned the analysis, said clinicians should be mindful of embedded bias in tools and take extensive family histories to get a fuller view of a patient’s risk for a disease.
“As users of these tools, clinicians should be aware that there’s very likely a bias ingrained, and while providers have little time with patients and are reliant on these tools, it’s about being mindful of those biases,” Escobar Alvarez said. “I’m hopeful the report brings clarity to an area where there’s been a lot of debate about what to do.”
When race correction was removed from the interpretation of results of the pulmonary function test, Black individuals showed a significantly higher prevalence and severity of lung disease. A recent study published in The New England Journal of Medicine found using the updated race-neutral Global Lung Function Initiative Global equations, which were developed in 2022 to improve equity in patient care, could reclassify pulmonary diagnoses of lung impairment for approximately 12.5 million patients in the United States.
However, implementation of the new equation has been slow and inconsistent, according to Escobar Alvarez. Many spirometers, for instance, still apply a correction for lung capacity of a 4%-6% decrease for Asian patients and a 10%-15% decreased lung capacity for Black individuals compared with White patients. Many health systems and clinicians are still using an outdated model for estimated glomerular filtration rate, despite the National Kidney Foundation updating its tool and guidelines in 2021.
The move away from using race incorrectly in tools and guidelines has been slowly growing, but the report is an accelerator, said Helen Burstin, MD, MPH, CEO of the Council of Medical Specialty Societies.
Burstin is leading the Encoding Equity in Clinical Research and Practice Initiative, which is in part funded by the Doris Duke Foundation. Launched this year, the project aims to identify inappropriate uses of race in algorithms and guidelines and redesign more accurate decision tools.
She pointed to the use of race in the risk calculator for miscarriage among women who previously had a cesarean delivery.
“We’ve known for years that Black women have a higher risk for hypertension,” Burstin said. “By putting race in that algorithm, they [researchers] missed the opportunity to realize it’s really about hypertension, and this is important for primary care providers because they can help patients with hypertension be well-controlled before they try to get pregnant.”
The updated vaginal birth after cesarean calculator was revised in 2021 to remove race and relies instead on variables like maternal age, body mass index, history of vaginal delivery, history of cesarean section, and gestational age. According to Burstin, the new calculator is still not widely used.
Instead of relying on race as a proxy, clinicians can focus more on travel background, environment, ancestry, and symptoms, said Genevieve Wojcik, MD, associate professor of epidemiology at the Johns Hopkins Bloomberg School of Public Health in Baltimore and an author of the report.
Wojcik noted that understanding patients can help inform clinicians on adverse health outcomes that might be spurred by structural racism, such as lower access to healthy foods or access to care.
“Just because race is a social construct doesn’t mean it’s not real; ignoring a person’s racial identity could also be harmful…it’s helpful for physicians to understand the communities they serve rather than making assumptions,” Wojcik said. “There need to be careful questions about what race represents; we tend to rely on it as a proxy for other things.”
Recommendations for Future Research from the National Academies of Sciences, Engineering, and Medicine
Researchers should provide definitions of race and ethnicity in manuscripts and other materials, if the information was self-reported or assigned, and which categories were used for enrollment or analysis and why.
Question the benefits of collecting race and ethnicity information if their use is not central to the research question.
Consider the use of other variables, such as immigration status, genetic markers, stress-related biomarkers, religion, and language, among others.
Researchers should not combine race and ethnicity categories with the purpose of improving statistical power.
The study was supported by contracts between the National Academy of Sciences, the Doris Duke Foundation, and the Burroughs Wellcome Fund.
Jennifer Lutz is a journalist based in New York and Barcelona, focusing on health, science, and policy. Get in touch at Jennifer-Lutz.com or @Jennifer_E_Lutz.
 
Send comments and news tips to [email protected].