ESPE Yearbook of Paediatric Endocrinology

To read the full abstract: Science 2019;366(6464):447-53. doi: 10.1126/science.aax2342

Short summary: Dissecting racial bias in health care systems revealed that a type of software program, which determines who receives access to high-risk health care management, routinely accepts healthier whites ahead of blacks who are less healthy.

Comment: The year 2020 will be remembered also by the death of George Floyd in police custody, which powered a global movement against racial injustice.

In the USA, large health systems and payers use algorithms to target patients with complex diseases to specially trained health providers. Such algorithms have been built to improve the care of patients with complex health, such as diabetes, by providing additional resources. Due to the high costs of the additional resources, health systems rely extensively on algorithms to identify patients who will benefit the most. While ‘the algorithms by themselves are neither good nor bad’, developers who build these algorithms rely on past data to design predictors of future health care needs.

The current study was set in a large academic hospital. All patients who were enrolled in a high-risk management program based on their health risk over a two-year period were identified and categorized according to their race. The current use of algorithms that determine who receives access to high-risk health care management programs was found to routinely accept healthier whites into the programs ahead of less healthy blacks. Using a different setting, an algorithm that specifically excluded race as a predictor more than doubled the number of black patients eligible to be enrolled in high risk management programs.

In other places in the world, algorithms in health care systems may also entail structural inequalities according to sex or ethnicity; these should be addressed and corrected.