SaO₂/FiO₂ ratio as risk stratification for patients with sepsis

Background

The PaO₂/FiO₂ ratio is a well-known parameter to assess respiratory dysfunction, used in Sequential Organ Failure Assessment (SOFA) [1]. This study aims to determine whether the SaO₂/FiO₂ ratio can be used in the assessment of respiratory impairment and as a predictor of ICU mortality in patients with sepsis and to evaluate its correlation with PaO₂/FiO₂.

Materials and methods

A retrospective cohort study conducted in the ICU of Hospital Santa Luzia, Brasilia, DF, Brazil, during 5 months. An arterial blood sample was collected at the time of admission. Patients with sepsis were divided into two groups: survivors group (SG) and nonsurvivors group (NSG). Correlation with SaO₂/FiO₂ and PaO₂/FiO₂ was evaluated with the Pearson correlation coefficient. Accuracy of SaO₂/FiO₂ and PaO₂/FiO₂ to predict ICU mortality was measured with the area under the receiver operating characteristic curve.

Results

A total of 118 patients with sepsis were enrolled. The mean age was 66 ± 21 years, SAPS3: 50 ± 14, APACHE II: 13 ± 8, PaO₂/FiO₂: 317 (IQ 233 to 426) and SaO₂/FiO₂: 362 (IQ 247 to 453). ICU mortality was 17.8% (n = 21). The main sites of infections were respiratory (57%, n = 67), urinary (19%, n = 23) and cutaneous (8.5%, n = 10). Nonsurvivor patients had lower SaO₂/FiO₂ (258 vs. 366, P = 0.00) and PaO₂/FiO₂ (285 vs. 354, P = 0.04). PaO₂/FiO₂ and SaO₂/FiO₂ had a good correlation (r = 0.645, P = 0.00). The relative risk of death in patients with SaO₂/FiO₂ <400 was 1.81 (95% CI: 1.47 to 2.24), SaO₂/FiO₂ <300 was 2.5 (95% CI: 1.54 to 4.05), SaO₂/FiO₂ <200 was 2.45 (95% CI: 1.27 to 4.71). The sensitivity for ICU mortality of SaO₂/FiO₂ <300 was 28% and of SaO₂/FiO₂ <200 was 35%. The specificity for ICU mortality of SaO₂/FiO₂ <300 was 90% and of SaO₂/FiO₂ <200 was 86% (95% CI: 93.5 to 100.0%). The area under the ROC curve for SaO₂/FiO₂ was 0.776 (95% CI: 0.677 to 0.875) and for PaO₂/FiO₂ was 0.655 (95% CI: 0.507 to 0.804) (Figure 1).

Figure 1

ROC curve for SaO ₂ /FiO ₂ and PaO ₂ /FiO ₂ .

Conclusions

A low SaO₂/FiO₂ was associated with mortality in this group of patients and had a good correlation with PaO₂/FiO₂. SaO₂/FiO₂ <300 showed high specificity for mortality. Further analysis is necessary to the validation of less invasive measures such as pulse oximetry saturation (SpO₂/FiO₂ ratio) in the assessment of patients with sepsis.