The Association of Skin Pigmentation With Pulse Oximetry Bias in Adult Patients in the ICU

Background Pulse oximeters often overestimate oxygen saturation in Black patients, and this measurement bias may contribute to worse outcomes, but the drivers of this observed bias need to be understood better. Research Question Does objectively measured skin pigmentation account for the racial bias in pulse oximeters in critically ill patients? Study Design and Methods We conducted a prospective observational cohort study of adult patients admitted to ICUs at 2 hospitals who underwent routine arterial blood gas (ABG) testing. Forehead skin pigmentation was measured using individual typology angle (ITA) and melanin index. Paired ABG-derived arterial oxygen saturation (Sao 2 ) and pulse oximetry oxygen saturation (Spo 2 ) within 5 minutes of each other were extracted from electronic medical records. Linear mixed-effects models were constructed to examine the association among race, skin pigmentation, and pulse oximetry bias (Spo 2 minus Sao 2 ). Results Among 198 patients with 709 paired measurements, Black race was associated significantly with pulse oximetry bias (unadjusted estimate, 1.032 [95% CI, 0.200-1.862]; P = .02), whereas ITA and melanin index (either as continuous or categorical variables) were not associated with pulse oximetry bias. In exploratory analyses, arterial pH was associated with pulse oximetry bias (estimate, –3.737 [95% CI, –6.281 to –1.029]; P = .04 after correction for multiple comparisons), whereas ITA and the interaction term between ITA and pH did not have significant associations. Interpretation In this study, skin pigmentation was not associated with pulse oximetry bias. Further investigation is required to understand additional factors that may cause the observed racial biases in pulse oximeter accuracy.