Scrutiny Continues Over Use of Race in Estimated GFR

/kidneynews/12_10_11/1/graphic/1f1.jpg

As calls for social justice and unrest reach into every corner of American life, the controversy over the inclusion of race as a factor in calculating estimated glomerular filtration rates reached more milestones as ASN and the National Kidney Foundation (NKF) formed a task force to reassess the practice, a congressional committee chair sought information from medical societies about the use of race in clinical algorithms, and more institutions moved away from the practice.

The issue even hit the mainstream media with a story from Consumer Reports, “Medical Algorithms Have a Race Problem.”

The NKF-ASN Task Force on Reassessing the Inclusion of Race in Diagnosing Kidney Disease is co-chaired by Cynthia Delgado, MD, and Neil R. Powe, MD, MPH, MBA. Delgado is an associate professor of medicine at the University of California, San Francisco, where Powe is also a professor of medicine.

Formed in August, the task force has begun its deliberations and plans to issue its initial recommendations by the end of 2020.

“We have been charged with examining the inclusion of race in the estimation of GFR and its implications for the diagnosis and subsequent management of patients with or at risk for kidney disease,” Delgado said. “We have also been charged with recognizing that any change in eGFR reporting must consider multiple social and clinical implications, be based on rigorous science, and be part of a national conversation about uniform reporting across healthcare systems. Those are just two of the five charges that we have gotten.”

The task force includes 14 members with broad expertise in healthcare disparities, epidemiology, health services research, genetic ancestry, clinical chemistry, patient safety and performance improvement, pharmacology, and social sciences, as well as two patients.

The recommendations will be vetted by an eGFR Advisory Board, Delgado said, and “go through a series of checks by members of the nephrology community at large, including patients and patient advocacy groups, to make sure that we are all in agreement with what we recommend.”

“We have been diligently working on this for months,” Delgado said. She noted that ASN has been playing a “key behind the scenes role on efforts to promote diversity, equity, and inclusion among kidney health professionals and to work toward eliminating health disparities in the communities that we serve.”

Congressional action

While significant racial disparities have long been present in the United States healthcare system, the dramatic racial health inequities caused by COVID-19 created a renewed focus and sense of urgency among ASN members to address and dismantle the systems inhibiting Black Americans from receiving equitable care. In May, ASN provided testimony to the House Committee on Ways and Means on the disproportionate impact of COVID-19 and kidney diseases on communities of color that highlighted dramatic racial health inequities in the US healthcare system.

On the heels of that hearing, the New England Journal of Medicine published “Hidden in Plain Sight—Reconsidering the Use of Race Correction in Clinical Algorithms,” which questioned the use of race in algorithms used in cardiology, nephrology, obstetrics, and urology. That spurred committee chair Richard E. Neal (D-MA) to write to ASN President Anupam Agarwal, MD, “to request an update about any work underway to investigate and change such clinical decisions support tools that fuel racial inequities in care.” Neal sent similar letters to other medical specialty organizations and requested a response by Sept. 25.

As part of the 7000-word response of behalf of ASN, Agarwal noted: “ASN agrees that unlike age, sex, and body weight, race is a social, not a biological, construct. Adjusting for race in these eGFR equations may not address the diversity within self-identified Black or African American patients as well as other racial or ethnic minority groups.”

The response further states: “ASN also recognizes that the causes of racial health inequities are multifactorial: disparities within kidney diseases are not limited to algorithms but have complexities that will need to be addressed beyond the NKF-ASN Task Force.”

In addition to her work on the task force, Delgado has been involved in responding to the request from the House committee. “I think this is a perfect opportunity for us to align ourselves with the House Ways and Means Committee so that we can focus on improving patient care. I think our response will create an opportunity for a dialogue,” she said.

How institutions are responding

Many of the colleagues Delgado has heard from around the country have expressed their eagerness to see the task force recommendations, but many hospital and academic centers are acting on their own to respond to concerns they are hearing from their own staffs about eGFR reporting practices. Beth Israel Deaconess Medical Center (BIDMC) was one of the earliest to make a change, in 2017, according to Melanie Hoenig, an associate professor of medicine at Harvard Medical School who spearheaded the effort at BIDMC.

It all started in the spring of 2016 during a class for medical students when she was talking about the African American coefficient that says that African Americans at a higher creatinine level can have the same eGFR as non-African Americans: “A medical student asked pointedly, ‘Why would there be a correction factor for a healthier value for the group at greatest risk of kidney disease?’”

That led to a literature review, meetings with people who helped create the formulas, and consultations with a variety of people, such as the head of the clinical laboratory and the chief of medicine. Hoenig said their investigation found “the use of race in clinical medicine is flawed and problematic. Ultimately, we agreed to change the language of the report to remove race but provide the two values generated by the eGFR formula.” The lab report indicates a range for the eGFR.

“Since race is a social construct and cannot be measured, use of race in the formula is fraught with problems and we feel that this practice should be abandoned. We hope that the joint task force from the ASN and NKF will also come to this conclusion,” Hoenig told Kidney News.

Task force co-chair Powe, who is also chief of medicine at Zuckerberg San Francisco General Hospital, said in contrast to the “well-vetted process” used at BIDMC, at his institution, “A small group dictated what happened.”

He said the change in the eGFR reports involved “changing the language and not the values themselves. [The new reports] just attach the label high muscle mass to African Americans and low muscle mass to other races.”

An alleged greater muscle mass among African Americans is often cited as the reason for higher creatinine levels, but Powe said: “People have been taught in medical school that the reasons creatinine levels are high have to do with muscle mass, but that is just a little dogma, and there are many other things that affect creatinine levels,” including tubular secretion, extra-kidney elimination, and the amount of meat in the diet. “The group that wanted to change to muscle mass now wants it changed to something else. They realized that what they initially did was not well thought out.” He fears that other institutions will fall victim to “advocacy efforts” rather than a thoughtful process of considering all the ramifications.

But the University of Washington, Vanderbilt University, and several hospitals affiliated with Brown University all simply dropped the use of the African American coefficient after an extended process. These institutions reported going through a careful process with input from many perspectives, but they also said that in the end it was not a difficult or controversial decision internally.

The change at the University of Washington came after a three-year process, according to Rajnish Mehrotra, MD, MS, interim head of the division of nephrology and editor-in-chief of CJASN. “The considerations that led us to the decision were the imprecision of the estimate itself, that a precise estimate is rarely imperative in clinical practice, and concern about using race as a biologic variable when it is actually a social construct,” Mehrotra said in an email to Kidney News.

“We presently report only a single value of eGFR using the CKD-EPI equation for all individuals without using the race coefficient. So, the value for non-blacks, as per the CKD-EPI, is reported for all individuals,” Mehrotra said. “The implementation has been widely accepted by all stakeholders, and I do believe the implementation has been successful. I credit the long time period over which we discussed this and the engagement with a broad range of stakeholders.”

Vanderbilt also dropped the African American reporting with “no replacement for race-related or adjusted reporting,” according to Alp Ikizler, MD, director of the division of nephrology, noting that ”race is not a reliable proxy for genetic difference and this is especially true in the US given the level of admixture that is present here.”

“The process required input from multiple parties including but not limited to leadership, clinical pathology, information technology, and most importantly clinicians. Once everyone was on board, the change was seamless,” he said. But it is too soon to judge the impact on patient care, and “we need to follow it up with quantitative [and] qualitative work to understand the impact of this change on Black people.”

Three teaching hospitals affiliated with his institution also dropped the use of race as a variable in estimating eGFR, said Douglas Shemin, MD, director of the division of kidney diseases and hypertension at the Alpert Medical School of Brown University. “We didn’t make this decision lightly,” Shemin said. We had people from general internal medicine, pathology, nephrology, kidney transplantation, etc., all involved. The decision was unanimous.”

“Our take-home point to our trainees is that they should look at the creatinine level in every single patient and make an assessment about what they think the actual glomerular filtration rate is based on clinical signs,” Shemin said. That assessment should include muscle mass and other potential factors.

The use of a race coefficient dates back to the Modification of Diet in Renal Disease (MDRD) study, in which all patients had their GFR measured by the kidney iothalamate clearance gold standard. The researchers concluded that African Americans had higher creatinine levels vis a vis a given GFR, leading to the use of a correction factor in the MDRD estimating equation.

Proponents of the value added by the equations tend to point to the MDRD and the evidence base of the equations and say the equations have served the nephrology community better than any alternative so far. But others question: How can you base a coefficient on “African American,” when, given the genetic and ancestral diversity of this population, you can’t define what it means? How reliable is an eGFR when one of the factors used—Black or not—is itself an estimate?

Even the terms of the debate can be muddled. “By dropping the race coefficient,” Powe said, “you are ignoring the data on African Americans that were in the study. It is saying, ‘Let’s just assign the [value of the] majority race. It is kind of saying, ‘White is right.’”

But Shemin considers the opposite to be the case, that using the coefficient singles out African Americans: “It is not African Americans vs. Caucasians; it is African Americans vs. anybody else.”

Andrew Levey, MD, chief of the division of nephrology at Tufts Medical Center who led much of the ground-breaking work on the estimating equations, said: “We always recognized that race was not the biological process by which African Americans differed from non-African Americans in the relationship between GFR and creatinine, but we thought it had too big of an effect to be overlooked. We understood that race was not a binary categorical variable, and that it wasn’t the biologic process, but that it stood in for something that was important.” The same consideration is true for age and sex, he said: “These are not the true biological variables, they are just easier to ascertain than the true biological variables.”

The 1999 MDRD study tried to identify the source of the variation with the statement “on average, black persons have higher muscle mass than white persons,” and the contentiousness of that statement has helped undermine confidence in the equations themselves considering that “neither the MDRD researchers nor any of the cited studies provided any evidence that Blacks indeed have higher muscle mass than whites,” wrote Vanessa Grubbs, MD, in a recent “Perspective” in CJASN. Shemin said that as part of his institution’s deliberations he reviewed the literature and could find no support for the assertion, nor any study that “looks at body weights and definitely shows that creatinine production is higher in blacks than whites.”

Just the same, Powe cautions against moving too quickly to discard a tool that has worked well: “What you hear the advocates of changing say is that these equations are hurting patients because they are leading to them not being on the wait list for a transplant, or being referred to a specialist. They cite anecdotal cases of individuals they may have taken care of. The advocates say that [the equations] caused harm, but there is no evidence of that other than the anecdotes,” Powe said. The disparities in care long pre-dated the use of the equations, and there is no evidence that the equations exacerbated them, he said. He worries that dropping the coefficient will lead to less accurate diagnoses and potential overtreatment among African Americans.

Richard Lafayette, MD, professor of nephrology at Stanford University Medical Center, agrees: “While a higher eGFR may bias toward waiting longer for transplant, it also biases toward being able to wait longer to start dialysis, waiting longer before meeting the worrisome diagnosis of chronic kidney disease, and a lower likelihood of being biased against by insurance. The key thing is to feel that the GFR prediction is accurate. The race correction is generally in the right direction by about the right amount, but can vary quite a bit individual by individual. Thus, a discussion should be undertaken with patients of any race, and other measures of kidney function can be considered.”

As the medical student in Hoenig’s anecdote might counter, as a population, the African American community is chronically undertreated. Overtreatment might be a preferred alternative, especially when early treatment can be beneficial when it comes to conditions like chronic kidney disease and transplantation.

But Powe notes that there is no evidence that the equations have anything to do with inequities that were in place before the equations existed: “If you want to truly address health inequities, let’s find the real drivers of them rather than scapegoating the equations.”

October-November 2020 (Vol. 12, Number 10 & 11)