Introduction: Management of children with congenital heart disease depends on accurate estimation of arterial oxygen saturation (SaO2). Two small single-center studies have suggested that pulse oximetry is less accurate in hypoxemic children of black race. We sought to confirm this relationship in a multicenter database. Methods: This retrospective cohort analysis extracted data from Cosmos, an Epic-specific database including over 1,700 hospitals. We included simultaneous arterial saturation (SaO2) and pulse oximetry (SpO2) readings within 2 minutes from encounters for children < 18 years old of black or white race (based on EHR demographics) with a cardiovascular complex chronic condition since 1/1/2020. To prevent misclassification of venous blood gases as arterial, we used the SaO2 with the smallest absolute SpO2-SaO2 for a given time point and excluded values where SaO2 was < 70%. We excluded children with multiple races, and those with only a patent ductus arteriosus. The cohort was described with summary statistics. SpO2-SaO2 differences were reported. Differences in SpO2-SaO2 by race were reported using Wilcoxon rank-sum tests and chi-square tests. Results: There were 22,512 included simultaneous SpO2/SaO2 readings across 9,301 encounters among 8,908 patients. The median (IQR) age was 3 (0-39) months. 9,812/22,512 (43.6%) of readings were on female patients, and 5,878/22,512 (24.8%) were on patients of black race. The median SaO2 readings were similar between black and white patients (96.0% 90.7%-98.8% vs. 96.2% 91.0-98.7%, p = 0.83). SpO2-SaO2 differences increased with progressive hypoxemia. The median SpO2-SaO2 difference for SaO2 readings≥95% was 0.0% (-1.3%, 1.0%). The median SpO2-SaO2 difference for SaO2 readings 80-84% was 8.0% (2.7%, 12.0%). Racial differences also increased with progressive hypoxemia; among SaO2 readings≥95%, 5.1% of white patients and 4.6% of black patients had SpO2-SaO2 differences≥5% (p = 0.32). However, among SaO2 readings 80-84%, 50.6% of white patients and 60.0% of black patients had SpO2-SaO2 differences≥5% (p < 0.001). Conclusions: Pulse oximetry bias and racial disparities increased with progressive hypoxemia. Clinicians should interpret pulse oximetry with caution in cyanotic congenital heart disease.
Fountaine et al. (Sun,) studied this question.