Clinician behavior when skin tone affects test results

Abstract

We compare care for Black vs. White patients following pulse oximetry, a widely used device for measuring blood oxygen levels which overestimates oxygen saturation in darker-skinned patients. Black patients are therefore medically more appropriate for follow-up care than White patients with the same pulse oximetry reading. Yet, using data from the Veterans Health Administration on 3.5 million emergency department visits between 2014 and 2018, we find that Black patients systematically receive lower rates of follow-up care than White patients with the same reading. Our findings illustrate how bias in a medical screening device can be propagated and potentially amplified in downstream care.