Removing Race Adjustment in Chronic Kidney Disease Care
A new study led by Stanford Health Policy researchers finds that algorithmic changes to a chronic kidney disease care equation are likely insufficient to achieve health equity as many other structural inequities remain.
Chronic kidney disease (CKD) affects more than one in seven adults in the United States, or about 37 million people. The burden of CKD is higher for Black or African American patients compared to white patients, as they’re three times more likely to progress to kidney failure, the final stage of CKD.
But the clinical indicator, the estimated glomerular filtration rate (eGFR) equation, commonly used to gauge CKD severity has for more than two decades included a “race adjustment” for Black or African American patients. The eGFR equation evaluates how efficiently a person’s kidneys are at filtering out creatinine, a waste product in urine.
The inclusion of race adjustment in the eGFR equation may have contributed to disparities in CKD and kidney failure outcomes through delays in nephrology referrals, preventative CKD treatment, and life-saving transplants. This led health equity experts, nephrologists, and the National Kidney Foundation to develop a new eGFR equation without race adjustment in an effort to remediate the propagation of racial bias in decision-making.
A new study led by Stanford Health Policy researchers finds that two years after implementation of the eGFR equation without race adjustment at Stanford Health Care did not result in changes to nephrology referrals or visits for Black or African American patients. They highlight that an overemphasis on further algorithmic changes should not divert attention from essential efforts to address social determinants of health and structural racism, key factors contributing to CKD disparities. The study has been accepted for publication in the Proceedings of the Conference on Health, Inference, and Learning.
“While a new eGFR equation without race adjustment could reduce disparities for Black or African American patients, it is important to acknowledge that decision-making is not guided by eGFR alone. Thus, it was critical to study the degree to which algorithmic changes to eGFR influenced decision-making and implications for health equity.” said Marika Cusick, a PhD student in health policy and lead author of the study. Cusick will present the work at the Conference on Health, Inference, and Learning this summer.
Testing the Theory
The Stanford researchers note in their study that, to their knowledge, there have been no reported prospective assessments of the impact on CKD care after implementation of the eGFR equation without race adjustment in clinical practice. So, they set out to do just that.
They analyzed more than 547,000 adult patients 21 and older who had at least one recorded creatinine test within the Stanford Health Care system from January 2019 through September 2023. Stanford Health Care implemented the updated eGFR equation on Dec. 1, 2021—without race adjustment. The researchers studied nephrology referrals and visits, which are critical for CKD management.
However, rates of referrals and visits demonstrated no change for all patients, including Black or African American patients. This study did not provide evidence for algorithmic changes leading to changes in decision-making for these outcomes. Instead, as demonstrated in prior literature, structural inequities, such as lower income status, lack of health insurance, reduced access to medical care, as well as poorer housing and environmental conditions are likely the key drivers of inequities in CKD.
“The inclusion of race adjustment in clinical equations can contribute to racial bias. However, it is essential that we recognize that changes to the eGFR equation are insufficient to tackle social factors and structural inequities,” said Sherri Rose, PhD, a professor of health policy and co-director of the Health Policy Data Science Lab, who is senior author of the study. “Amidst increasing hype around AI, focusing too heavily on algorithms should not redirect crucial funding and resources from research and interventions that aim to tackle structural causes of health and health care disparities in kidney disease.”
The researchers concluded: algorithmic changes will not be enough.
The additional coauthors of the paper are Glenn M. Chertow, MD, MPH, professor of medicine (nephrology) and associate chair of the Department of Medicine; Douglas K. Owens, MD, professor of health policy and chair of the Department of Health Policy; and Michelle Y. Williams, PhD, RN, associate chief nursing officer, research & health equity at Stanford Health Care and clinical assistant professor of medicine.
This research was funded by Stanford Impact Labs, which led to the launch of the Health Care Fairness Impact Lab.
