The Search for Optimal Oxygen Saturation Targets in Sepsis Patients: a retrospective study

Abstract

Abstract Background Hypoxemia or hyperoxia oxygenation might increase in-hospital mortality in patients with sepsis. However, the optimal oxygenation target remained unknown. We performed a study to explore and validate the optimal oxygen saturation targets in sepsis patients. Methods Medical records of sepsis patients from 2008 to 2019 were retrieved from the Medical Information Mart for Intensive Care (MIMIC)-IV database. A generalized additive model (GAM) was established to comprehensively analyze the U-shaped nonlinear relationship between in-hospital mortality and time-weighted average (TWA)-SpO2 during oxygenation. The lowest and flattest area was identified as the optimal oxygenation target. A logistic regression was utilized to verify the optimal oxygenation target. Finally, a mediation analysis was adopted to estimate the direct effect of the TWA-SpO2 level by excluding the indirect effect of the fraction of inspired O2(FiO2). Results 3800 sepsis patients with 630,704 SpO2 records were enrolled in our study, among whom 705 (18.5%) passed away at hospital discharge. The median age of the patients was 54.31 [52.22, 76.66] years, and 2326 (61.21%) were male patients. The GAM analysis demonstrated a U-shaped nonlinear relationship between TWA-SpO2 and in-hospital mortality during oxygenation. The SpO2 interval corresponding to the curve's lowest region of in-hospital mortality was 95%~98%. In accordance with the relationship between the TWA-SpO2 and the target interval range, the patients were segregated into hypoxemia, hyperoxia, and optimal oxygenation group, respectively. The multivariate logistic regression results also presented that the in-hospital mortality risk in the hypoxia oxygenation groups increased by 60.4% (OR=1.60, 95%CI 1.23~2.09, P<0.001), while patients in the hyperoxia oxygenation group also increased by 61.3% (OR=1.61, 95 %CI: 1.26-2.06, P<0.001), by adopting the sepsis patients of the optimal oxygenation group as the control. Finally, the mediation effect analysis results presented that the direct effect of the TWA-SpO2 was 0.072 (95% CI: 0.030-0.110, P<0.001). Conclusions The optimal pulse oximetry targets for oxygenation in sepsis patients were likely 95%~98%, corresponding with the lowest risk of in-hospital mortality. It was also observed after the indirect effect of FiO2 was removed. Trial registration: Our study was registered with the Chinese Clinical Trials Registry on January 11, 2021, where the registration number was ChiCTR2000040149. http://www.chictr.org.cn/edit.aspx?pid=64427&htm=4.