Should physicians reconsider the use of race correction in clinical algorithms?
Click Here to Manage Email Alerts
Yes, clinicians should consider other important variables apart from race.
I first entered this debate in the context of HbA1c. HbA1c is commonly used as a measure of average glycemia, an indicator of risk for the development of diabetic complications and an index of the quality of diabetes care. Observational studies have consistently demonstrated that, for any level of glycemia, Black adults have higher HbA1c levels compared with white adults. That difference is not readily explained by differences in demographics, physical or physiologic parameters such as age, BMI, insulin resistance, or access to or quality of care. More recently, we found that genetic polymorphisms that operate through nonglycemic mechanisms are associated with racial difference in HbA1c, although, taken together, these polymorphisms account for a small proportion of the observed difference. On the other hand, a relatively large proportion of the difference in HbA1c levels between whites and Blacks is explained by genetic markers of African ancestry, suggesting that other yet-to-be-discovered genes or nongenetic factors associated with being Black may account for the difference.
Failure to recognize and account for racial differences in HbA1c could potentially result in harmful overtreatment among Black adults, which could result in excess hypoglycemia. At the same time, increasing HbA1c targets for Blacks by using a race correction could result in undertreatment and increased rates of microvascular and neuropathic complications.
Rigorous scientific investigation should be undertaken to investigate the reasons for the observed difference between racial groups and those studies should comprehensively address demographic, physical, physiological, and genetic parameters, as well as health care access and quality. A recent review has also highlighted the importance of social determinants of health in diabetes. We must also consider the impact of education, income, economic opportunity, housing, transportation, neighborhood, physical environment, and systemic racism as potential explanations for observed racial differences in HbA1c. A person’s ZIP code may turn out to be the most important nonglycemic determinant of HbA1c. It is important to identify the factors that truly account for racial differences in HbA1c so that root causes can be understood and addressed.
William H. Herman, MD, MPH, is professor of endocrinology and epidemiology at the University of Michigan. Disclosure: Herman reports he serves on a data safety monitoring board for Merck.
No, algorithms are not the drivers of racial disparities.
The use of race in algorithms for clinical care, including for kidney disease, is generating discourse and action about systemic discrimination in health care. Recently, several institutions have taken steps to remove the use of race in equations involving eGFRs. Many of the institutions dropping this coefficient did just that — simply dropped it, without an adequate replacement.
The effect of removing race on clinical decisions is unknown and is likely to be profound, affecting everything from the administration (eg, metformin) and dosing of medications, to considering kidney donation and research study participation for new therapies, any of which might exacerbate existing health disparities or impede research with diverse participants. A recent study showed that the FDA’s label change in 2016 for metformin from creatinine to race equations eliminated a disparity in Blacks receiving metformin (Shin JI, et al. J Am Soc Nephrol. 2020; doi:10.1681/ASN.2019101119).
There may be regulatory, public health and economic considerations. Studies supporting approval of drugs currently under consideration by the FDA use the eGFR equations. Government agencies incorporate the equations in their long-standing efforts to track kidney disease. Assigning a diagnosis of kidney disease to Black patients could affect their ability to secure life insurance. Clearly, all the ramifications and long-term health consequences are not being considered, estimated or measured.
We need to slow down. We have had these race equations for 20 years. There is conjecture that, at least in nephrology, such an equation decreases access to specialists and decreases access to the transplant waiting list. All of that, however, is either anecdotal or thought experiments — not evidence. Racial disparities existed and were documented as early as the 1980s, decades before a race equation was used, through the 1990s and persisting today. Therefore, the algorithm did not cause the disparities. Hopefully, we will see more evidence of the harms and benefits of these types of equations before large-scale changes are made. If anything, things have gotten better with respect to transplant waiting lists for Black adults. Race equations have become a scapegoat, and in the end, eliminating them could ultimately cause more harm. This movement is well intentioned, but the implementation is flawed.
Neil R. Powe, MD, MPH, MBA, is the Constance B. Wofsy Distinguished Professor and vice chair of medicine at the University of California, San Francisco, and chief of medicine at the Priscilla Chan and Mark Zuckerberg San Francisco General Hospital. Disclosure: Powe reports no relevant financial disclosures.
Collapse
Read more about
Click Here to Manage Email Alerts