Abstract Rationale Hemoglobin (Hb) concentration plays a central role in oxygen delivery and has been hypothesized to influence oxygen saturation measurements, including arterial oxygen saturation (SaO2) and pulse oximetry (SpO2). However, the extent to which hemoglobin independently affects SpO2 or SaO2 in acid-base conditions remains unclear. Objectives To determine whether hemoglobin concentration is associated with SpO2 or SaO2 in non-acidemic ICU patients and to evaluate whether this association differs when defining non-acidemia using two clinically relevant pH cutoffs: ≥7.30 and ≥7.35. Methods We performed a retrospective analysis of adult ICU encounters in the BOLD database. Both conservative and liberal acidemia approaches were defined using two alternative thresholds: (1) pH 7.30, pH 7.30-7.45 and pH 7.45 (2) pH 7.35, pH 7.35-7.45, and pH 7.45. Primary outcomes were SpO2 and SaO2 (continuous). Predictors included hemoglobin concentration, modeled both continuously and categorically (9 g/dL and 9 g/dL). Statistical analyses included Spearman’s correlation coefficient (rho), sensitivity modeling of hemoglobin levels below and above 9 g/dL were performed with pulse/arterial oximetry. Results The mean difference between SpO2-SaO2 is higher in acidemia 1.44 (pH 7.35) to 1.50 (pH 7.30), and 1.12 (pH 7.30-7.45) to 1.21 (pH 7.35-7.45), versus 0.82 in alkalosis (pH 7.45). Across both definitions of non-acidemia and acidemia, hemoglobin showed no statistically significant correlation with SaO2, however showed significant negative correlation with SpO2 (rho -0.08, p 0.001) regardless of acid-base disorder, and with SaO2 (rho -0.028, p = 0.008) in acidemia only. The variability among SaO2-SpO22 is more marked in acidemia as compared to patients with alkalosis or without acid-base disorder. Thesefindings remained stable across subgroups defined by different pH cut-offs for acidemia. In contrast, expected physiologic determinants such as PaCO2 (rho -0.166 vs -0.150) and pH (rho= 0.134 vs 0.086) remained strongly associated with SaO2 as compared to SpO2, supporting internal validity of the models. No evidence was observed that anemia (Hb 9 g/dL) altered SpO2-SaO2 concordance in non-acidemic vs acidemia states. However, more variability in SpO2-SaO2 was observed when Hb 9g/dL. Conclusions In this large multicenter ICU cohort, hemoglobin concentration was only associated with variability in SaO2 among patients with acidemia, regardless of whether non-acidemia was defined by pH ≥ 7.30 or ≥ 7.35. These findings suggest that, in the absence of acidemia, hemoglobin does not meaningfully influence oxygen saturation measurements or the SpO2-SaO2 relationship. This clarifies the physiologic conditions under which hemoglobin affects oxygenation metrics and supports the robustness of saturation-based monitoring in acidemic patients. This abstract is funded by: N/A
Yousaf et al. (Fri,) studied this question.