Acidemia, Lactate, and Anemia Are Associated With Clinically Meaningful SpO2-SaO2 Discordance in Critically Ill Adults.

BACKGROUND: Pulse oximetry is central to ICU care, yet clinically meaningful SpO₂-SaO₂ discordance is common and may reflect both racial/ethnic disparities and physiologic derangements. The independent contributions of acidemia, hyperlactatemia, and anemia to discordance have not been comprehensively evaluated in a large, diverse ICU cohort. METHODS: We performed a retrospective cohort study using the Blood-gas and Oximetry Linked Dataset (BOLD). We analyzed patient-level paired SpO₂ and SaO₂ measurements within 5 minutes, limiting to one pair per hospitalization. After exclusions and complete-case requirements for hemoglobin, pH, lactate, race/ethnicity, and prespecified covariates including SOFA (past 24 hours), MAP, and temperature, the primary analytic cohort included 10,094 adults. Outcomes were signed discrepancy (ΔSat = SpO₂ - SaO₂), clinically significant discordance (|ΔSat| ≥4%), and directional misclassification (underestimation ≤-4; overestimation ≥+4). Multivariable OLS, logistic, and multinomial regression were used; pH × lactate interaction was tested using mean-centered terms. RESULTS: Mean ΔSat was 0.93±3.74%, and 18.5% met |ΔSat| ≥4%. Lower pH, higher lactate, and lower hemoglobin were independently associated with discordance. Black race and Hispanic ethnicity were associated with greater overestimation versus White patients after adjustment. A significant pH × lactate interaction suggested attenuation of the lactate-ΔSat association at lower pH. Findings were consistent in a larger sensitivity cohort (N=18,804). CONCLUSIONS: In critically ill adults, acidemia, hyperlactatemia, and anemia are associated with clinically meaningful SpO₂-SaO₂ discordance, alongside persistent racial/ethnic differences in overestimation. Clinicians should interpret SpO₂ cautiously in metabolically stressed and anemic patients and consider confirmatory testing when clinical concern exists.