[Exploring the optimal range of pulse oxygen saturation in patients with sepsis: a retrospective study based on MIMIC- IV data].

Objective: To explore the optimal pulse oxygen saturation (SpO₂) range during hospitalization for patients with sepsis.

Methods: A case-control study design was employed. Demographic information, vital signs, comorbidities, laboratory parameters, critical illness scores, clinical treatment information, and clinical outcomes of sepsis patients were extracted from the Medical Information Mart for Intensive Care- IV (MIMIC- IV). A generalized additive model (GAM) combined with a Loess smoothing function was employed to analyze and visualize the nonlinear relationship between SpO₂ levels during hospitalization and in-hospital all-cause mortality. The optimal range of SpO₂ was determined, and Logistic regression model along with Kaplan-Meier curve were utilized to validate the association between the determined range of SpO₂ and in-hospital all-cause mortality.

Results: A total of 5 937 patients met the inclusion criteria, among whom 1 191 (20.1%) died during hospitalization. GAM analysis revealed a nonlinear and U-shaped relationship between SpO₂ levels and in-hospital all-cause mortality among sepsis patients during hospitalization. Multivariable Logistic regression analysis further confirmed that patients with SpO₂ levels between 0.96 and 0.98 during hospitalization had a decreased mortality compared to those with SpO₂ < 0.96 [hypoxia group; odds ratio (OR) = 2.659, 95% confidence interval (95%CI) was 2.190-3.229, P < 0.001] and SpO₂ > 0.98 (hyperoxia group; OR = 1.594, 95%CI was 1.337-1.900, P < 0.001). Kaplan-Meier survival curve showed that patients with SpO₂ between 0.96 and 0.98 during hospitalization had a higher probability of survival than those patient with SpO₂ < 0.96 and SpO₂ > 0.98 (Log-Rank test: χ ² = 113.400, P < 0.001). Sensitivity analyses demonstrated that, with the exception of subgroups with smaller sample sizes, across the strata of age, gender, body mass index (BMI), admission type, race, heart rate, systolic blood pressure, diastolic blood pressure, mean arterial pressure, respiratory rate, body temperature, myocardial infarction, congestive heart failure, cerebrovascular disease, chronic liver disease, diabetes mellitus, sequential organ failure assessment (SOFA), simplified acute physiology score II (SAPS II), systemic inflammatory response syndrome score (SIRS), and Glasgow coma score (GCS), the mortality of patients with SpO₂ between 0.96 and 0.98 was significantly lower than those of patients with SpO₂ < 0.96 and SpO₂ > 0.98.

Conclusions: During hospitalization, the level of SpO₂ among sepsis patients exhibits a U-shaped relationship with in-hospital all-cause mortality, indicating that heightened and diminished oxygen levels are both associated with increased mortality risk. The optimal SpO₂ range is determined to be between 0.96 and 0.98.