Health algorithm biased against black individuals finds study

By Dr. Ananya Mandal, MDOct 26 2019

A new study has revealed that a commonly and widely used healthcare algorithm may be biased against blacks and may be denying them of vital medical treatment. The study titled, “Dissecting racial bias in an algorithm used to manage the health of populations,” was published this week in the latest issue of the journal Science.

Image Credit: SeaRick1 / Shutterstock

Lead author Ziad Obermeyer, an acting associate professor at the University of California, Berkeley says without naming the algorithm that “almost every large health care system” at present uses this algorithm for their treatment and this includes major institutions as well as insurers. He said, “This is a systematic feature of the way pretty much everyone in the space approaches this problem.” According to the study over 200 million people and their healthcare come under the umbrella of this algorithm.

The authors of the study explained that most health care providers use this algorithm to screen patients and triage them for high risk medical care and interventions. Those with complex comorbidities and serious illnesses are selected by the algorithm for treatment and intervention. These individuals are ear marked and may receive more attention from the health care providers. These individuals get priority appointments with doctors and get dedicated nursing care. For the health care system, explained the authors of the study, this kind of algorithm saves millions and also ensures that the ones that most need the service and catered to first. This ensures the best and optimum utilization of resources.

In order to select these patients, the algorithm utilizes cost data according to each of the patients and this gives them a fair idea about the need of the patient. According to the authors of the study, their research finds that black individuals have poorer access to health care in the first place. This means that less is spent on their health. The algorithm might be thus missing these individuals as they are not among the patients that are more costly to the healthcare system. The team explains that the discrepancy of unequal access to healthcare thus is not addressed by this universally used and accepted algorithm. Obermeyer said, “Cost is a reasonable proxy for health, but it’s a biased one, and that choice is actually what introduces bias into the algorithm.”

Sendhil Mullainathan, a professor of computation and behavioral science at the University of Chicago Booth School of Business and author of the study, said in a statement, “It’s truly inconceivable to me that anyone else’s algorithm doesn’t suffer from this. I’m hopeful that this causes the entire industry to say, ‘Oh, my, we’ve got to fix this.’ ”

According to their findings, the team noted that at present the additional healthcare service and attention is received by 17.7 percent of the black population. If the baseline costs could be brought at par with the white population, 46.5 percent of the black patients would need to be on the special care list. The team looked at 3.7 million individuals and how the algorithm classified them for their study. They found that black spent $1,800 less in medical costs per year than white patients when both individuals had same number of chronic diseases.

According to Obermeyer, there are many disparities between a white individual and a black person in terms of health care spending. Loss of work, baseline poverty, pay cuts due to health grounds are all contributors to the disparity he said. He added, “There are just a million ways in which poverty makes it difficult to access health care.” Bias in terms of doctors treating black patients also comes into play he said.

Obermeyer added however that this problem is not unsolvable and could be easily remedied. He said, “That bias is fixable, not with new data, not with a new, fancier kind of neural network, but actually just by changing the thing that the algorithm is supposed to predict.” He said that the algorithm needs to be focussing on health outcomes of all patients rather than cost alone. This could solve this disparity problem. He concluded, “With that careful attention to how we train algorithms we can get a lot of their benefits, but minimize the risk of bias.”

In an accompanying article with the study, Ruha Benjamin, an associate professor of African American studies at Princeton University, spoke about the story of Henrietta Lacks. Lacks was a young African American mother with cervical cancer whose cervical cells well used for biomedical research without her knowledge or consent. “I am struck by how many people still think that racism always has to be intentional and fueled by malice. They don’t want to admit the racist effects of technology unless they can pinpoint the bigoted boogeyman behind the screen,” Benjamin said.