By Mitchell Linder, MD
Assistant Professor, Department of Obstetrics and Gynecology, University of Rochester School of Medicine
and Dentistry, Strong Memorial Hospital, Rochester, NY
Dr. Linder reports no financial relationships relevant to this field of study.
SYNOPSIS: Many clinical calculators use race as a predictive variable to assess risk for outcomes. Although most of the tools assume a genetic disposition for these outcomes, other factors, such as health disparities and other potential confounders, are more likely to be the underlying reasons for any race-related differences in outcomes.
SOURCE: Vyas DA, Eisenstein LG, Jones DS. Hidden in plain sight – reconsidering the use of race correction in clinical algorithms. N Engl J Med 2020;383:874-882.
This is a theoretical analysis evaluating multiple existing diagnostic algorithms and clinical prediction guidelines and their use of race or ethnicity categories as inputs for calculation. The authors identified 13 prominent tests and calculators in wide use today that each use race/ethnicity as a component in their formulas. Studies they identified include pulmonary function testing, the Fracture Risk Assessment tool (FRAX), the Simplified Calculated Osteoporosis Risk Estimation (SCORE), the Breast Cancer Surveillance Consortium Risk Calculator, the National Cancer Institute Breast Cancer Risk Assessment Tool, the Rectal Cancer Survival Calculator, the pediatric urinary tract infection calculator (UTICalc), the STONE score (used for prediction of possible kidney stones), the Vaginal Birth After Cesarean (VBAC) risk calculator, the Kidney Donor Risk Index, the estimated glomerular filtration rate (eGFR) calculator, The Society of Thoracic Surgeons Short-Term Risk Calculator, and The American Heart Association’s Get with the Guidelines — Heart Failure.
They noted that this list is not all-inclusive but includes readily available examples about how pervasive the use of race/ethnicity is in medical decision-making tools.
The authors examined each of these by specialty to show how each equation uses “race-correction” and how each of these race adjustments potentially could negatively affect Black patients. In cardiology, the heart failure score predicts lower risks for Black patients (without having any rationale provided), and the authors noted this could cause clinicians and hospitals to devote less resources to these patients, since they are deemed lower risk. For nephrology, they specifically pointed out how calculators overestimate kidney function based on assumptions about race-related differences in creatinine and overestimate the potential for kidney transplant failure involving kidneys from Black donors. The STONE score also is noted to assign lower scores to Black patients, thereby potentially guiding clinicians away from a possible diagnosis. The VBAC calculator is noted to predict a lower chance of success for African American and Hispanic patients. This decreased predicted success rate is hypothesized to discourage these patients from an attempt at a trial of labor and, thus, further increase the disproportionately high rate of cesarean delivery that minorities experience already.
The authors urged a thorough review of existing tools by institutions, medical societies, and individual clinicians. They noted that to do this properly involves working to re-evaluate how clinicians conceptualize race and apply it to the care they provide.
COMMENTARY
Systemic racism abounds in the medical field, with clinical calculators as just one example where it can be seen. Past research has shown that commonly employed commercial prediction algorithms used to allocate healthcare also have underlying racial bias in how their calculations are made.1 Unfortunately, we are not able to look under the hood to see the inner workings of the tools we often use. How, then, are clinicians to go about making evidence-based clinical decisions that incorporate guideline-supported rationale? The authors provided three novel criteria that clinicians should consider: “When developing or applying clinical algorithms, physicians should ask three questions: 1) Is the need for race correction based on robust evidence and statistical analyses (e.g., with consideration of internal and external validity, potential confounders, and bias)?; 2) Is there a plausible causal mechanism for the racial difference that justifies the race correction?; and 3) Would implementing this race correction relieve or exacerbate health inequities?”
Use of these types of questions could be helpful in identifying inherent bias and systemic racism that potentially may be harming our patients because of the automatic assumption that clinical assessment tools are de facto free of prejudice. It is the responsibility of each and every practitioner to provide unbiased and appropriate care for our patients. Making sure our clinical tools are based on these same principles is a good start. For example, as of June 1, 2020, the University of Washington labs have transitioned from using the eGFR formula to using one that excludes race as a variable.2-4
The inclusion of the VBAC calculator on this list is especially concerning for clinicians who use this score as their primary determinant for providing delivery option guidance to their patients. Analysis of this calculator specifically has shown that most of the race-related statistics that underpin the algorithm likely are related to social advantage or disadvantage, as opposed to being the result of any specific race.5 Given these known assumptions that are built into the VBAC calculator, use of the tool becomes more dangerous, since it could serve to perpetuate health disparities by having potentially viable trial of labor after cesarean candidates steered to repeat cesarean delivery solely because of their race.
With the higher rate of maternal morbidity and mortality that minorities sustain in the United States and the world, it is our duty to look at the tools we use, such as the VBAC calculator, to determine what potential conflicts or biases are included and whether that opinion is valid and applicable to our patients.6 Of course, that one step will not reverse entrenched biases everywhere nor eliminate systemic racism in one sweep, but every act to combat inequalities is a worthwhile one to undertake.
REFERENCES
- Obermeyer Z, Powers B, Vogeli C, Mullainathan S. Dissecting racial bias in an algorithm used to manage the health of populations. Science 2019;366:447-453.
- Eneanya ND, Yang W, Reese PP. Reconsidering the consequences of using race to estimate kidney function. JAMA 2019;322:113-114.
- Bonham VL, Green ED, Pérez-Stable EJ. Examining how race, ethnicity, and ancestry data are used in biomedical research. JAMA 2018;320:1533-1534.
- University of Washington. UW Medicine to exclude race from calculation of eGFR (measure of kidney function). May 29, 2020. medicine.uw.edu/news/uw-medicine-exclude-race-calculation-egfr-measure-kidney-function
- Vyas DA, Jones DS, Meadows AR, et al. Challenging the use of race in the vaginal birth after cesarean section calculator. Womens Health Issues 2019;29:201-204.
- Petersen EE, Davis NL, Goodman D, et al. Racial/ethnic disparities in pregnancy-related deaths — United States, 2007-2016. MMWR Morb Mortal Wkly Rep 2019;68:762-765.
