How a racist algorithm kept Black patients from getting kidney transplants
In recent years, doctors have started to lean on algorithms when determining certain aspects of patient treatments. While these systems are designed to take into account information like medical test results and risk factors facing certain people, they have also exacerbated existing biases against Black patients. A new study published this month in the Journal of General Internal Medicine found that one algorithm used for assessing a person's kidney function often overestimated the health of Black patients, resulting in them receiving less specialized care, or worse, keeping them from getting placed on a kidney transplant waitlist.
The study looked at the health records of 56,845 people dealing with chronic kidney disease in the Mass General Brigham health system in Boston, including 2,225 Black patients. By analyzing these records, researchers found that more than 700 Black people suffering from kidney disease were given healthier scores than their white counterparts despite displaying the same conditions and risks. In other words, one in three Black people treated within the healthcare network were misdiagnosed, and would have been placed in a more severe category of kidney disease had they been white.
The effects of this type of misdiagnosis can be significant. In most cases, those patients would have been referred to a kidney specialist in order to receive specialized treatment for their condition. In 64 cases, or about 10 percent of those who had their kidney health miscalculated, recalculating would have produced a result qualifying them to be placed on a kidney transplant list. None of those patients were handed off to a specialist or evaluated for a transplant, suggesting the doctors never questioned the algorithm's results.
The system is designed to calculate CKD-EPI, which is a measurement of kidney function based on blood testing. The lower the score, the worse a person's kidney is functioning. That score is then used to determine how severe the patient's kidney disease is and what level of treatment they require. Black patients, the study found, had their scores artificially inflated by 15.9 percent.
The kidney score algorithm is far from the only tool using race in its decision-making. A study published in the New England Journal of Medicine this summer found more than a dozen algorithms that make use of a patient's race in determining treatment. The tools range from helping to make treatment decisions for heart disease to cancer diagnosis.
These systems all account for Black patients differently but produce a similar result: codifying and automating a long history of worse health treatment for Black people. Sometimes Black patients are dinged for their socioeconomic status. A study published last year in the journal Science found that an algorithm that used health costs as a proxy for health needs falsely concluded that Black patients are healthier than white patients. In some cases, the algorithms trade on false and gross stereotypes. That was the case according to a lawsuit filed against the NFL, which claimed the league used an algorithm that Black players started with worse cognitive functioning than white players.
Regardless of the reason, these algorithms are resulting in Black patients getting worse care, a problem that has plagued the US healthcare system for decades. Studies have found that doctors have implicit racial biases that result in Black patients getting less or worse care, from failing to provide pain medication to misdiagnosing mental illnesses. Some doctors and health care facilities have already given the boot to algorithms that include race in calculations, and lawmakers are calling for increased scrutiny into these systems. But much work needs to be done to remove the bias against Black patients, as the algorithms simply systematize the decisions previously made by health professionals.