The Utilization of Race-Based Clinical Algorithms to Make Healthcare Decisions
Do race-based clinical algorithms undermine health equity? In 2023, PlusInc published a blog entitled “Diversity in Clinical Trials Still Lags,” which focused on the various issues related to a lack of gender, racial, and ethnic diversity in clinical and medical device trials, highlighting one particularly galling disparity that was of major importance during the COVID-19 global pandemic:
“A retrospective cohort study released in the Journal of the American Medical Association Internal Medicine found that compared to white patients, “…Asian, Black, and Hispanic patients had a higher adjusted time-weighted average pulse oximetry reading and were administered significantly less supplemental oxygen for a given hemoglobin oxygen saturation compared with white patients” (Gottlieb, et al., 2022).”
The 2023 blog post focused on how that lack of inclusion of women and racial and ethnic minorities in clinical and medical device trials can result in the development of vaccines, medications, and devices that inadvertently presuppose that the patient being treated is White, potentially resulting in worse diagnostic and health outcomes for Black, Indigenous, and other People of Color (BIPOC). This blog focuses on another aspect of healthcare that may be resulting in significant diagnostic and health outcomes: race-based clinical algorithms.
This year, a recent article published as part of STAT’s “Embedded Bias” series shed light on another issue that is both deeply embedded in clinical practice and is potentially exacerbating negative health outcomes among minority populations: the use of race-based clinical algorithms to determine care pathways.
In their article, Katie Palmer and Usha Lee McFarling found that clinical algorithms—a series of questions or steps, in some cases a flow-chart, ostensibly designed to assist providers with making medical decisions about a patient—continue to use outdated and biologically unsupported assumptions based upon race (Palmer & McFarling, 2024).
One such assumption inaccurately estimated a higher level of kidney function in Black patients than for non-Black patients with otherwise identical patient-specific variables. This Estimated Glomerular Filtration Rate (eGFR) value measures how quickly a person’s kidneys remove creatinine from the blood, and the inclusion of race as a factor in that measurement has resulted in Black patients being diagnosed with kidney failure at much later stages in the disease compared to non-Black patients causing them to miss out on years of vital dialysis treatment, exacerbating co-morbid conditions, and delaying their placement on kidney transplant lists (United Network for Organ Sharing, 2023). And this race-based algorithm has been in place since 1999 (National Institutes of Health, 2021).
The use of a race coefficient in this coefficient for the eGFR test traces back to a 1998 study that showed that Black patients had higher serum creatinine levels, on average, than White patients, but didn’t examine why some groups may have higher levels compared to others. These higher creatinine levels were associated with better kidney function in Black patients than was actually the case, and researchers found that, when using the 2009 CKD-EPI eGFR formula, all Black patients automatically received a 16% higher score in their eGFR test results because of this flawed, race-based assumption. To put this into greater perspective, Black patients have a higher rate of chronic kidney disease than their white counterparts, being more than three times as likely to develop kidney failure (Nicholas, 2022).
Thankfully for Black patients, the Board of Directors of the Organ Procurement and Transplantation Network (OPTN) unanimously approved a process to backdate the waiting times of Black kidney transplant candidates who were disadvantages by the previous use of the race-based eGFR calculation, the use of which has been prohibited since July 27th, 2022 (OPTN, 2023).
Other race-based algorithms, however, are more firmly entrenched. As Palmer and McFarling found, the American healthcare system is both unaccustomed, and often hostile, to reassess, replace, and discontinue long-held assumptions, beliefs, and practices when it comes to patient care (Palmer & McFarling, 2024). This is in part because the American healthcare system is neither centralized, nor designed to quickly adapt to changes due to being a for-profit model designed to maximize profits at both the provider and payor levels while minimizing expenditures on patient care by payors, and in part because clinicians simply don’t want to admit being wrong.
The disjointed and disconnected nature of the healthcare system—by which we mean the decentralized nature of a system that spans well over 3,000 miles from the furthest Eastern point to the furthest Western point of the United States—makes implementing changes to clinical algorithms on a broad scale incredibly difficult. While some providers living in urban and suburban areas may be quicker to adopt new algorithms that remove race-based assumptions from their calculations, others will be slower to adapt. A good way to think of it is to look at the adoption of broadband (and even fiberoptic) Internet service: while large cities were quick to transition from dial-up Internet services using telephone wires to broadband services using coaxial and ethernet cables, it took smaller cities, suburbs, and rural areas considerably longer to make this transition. For some, even basic broadband and Wi-Fi services that many of us take for granted simple don’t exist in parts of the United States. The same is true for changes in healthcare standards.
In addition to the slow adoption of and resistance to replacing race-based algorithms, there is little appetite for enforcing such changes. While a new rule that will prohibit discrimination through the use of patient care decision support tools is set to go into effect in May 2025, the U.S. Department of Health and Human Services constructed the rule in such a way as to support voluntary compliance, rather than mandated and enforced compliance (Palmer & McFarling, 2024). Moreover, as we’ve seen with other aspects of healthcare provision, DHS has consistently been hesitant to enforce compliance with its rules, either through sanctions, fines, or civil or criminal charges. Because physicians are licensed by different administrative and governmental bodies, there is no centralized control over who can and cannot practice medicine in the United States, meaning that physicians who fail to adhere to basic standards of care and continue to use tools that result in discriminatory and race-based diagnoses and treatments face limited scrutiny or consequences for failing to comport with new standards.
Sadly, these kinds of changes take far too long to take hold in the American healthcare system. But, theycanhappen. The COVID-19 pandemic demonstrated how quickly a system can adapt if sufficient technical assistance, supportive, and financial resources are provided to ensure that new protocols are adopted. The reality, however, is that we need to move faster to ensure that non-White patients are not being inadvertently harmed because our healthcare system is stuck in an inertial state where doing “what we’ve always done” is easier—and let’s face it, cheaper—than identifying, diagnosing, and fixing the problems that exist