375

Abstract

Introduction: Hyperoxia with excessive supplemental oxygen (O2) may be injurious. While several studies have provided epidemiological data for oxygen administration in critically ill patients, evidence is inadequate for deciding the best target values for arterial O2 tension (PaO2). Methods: We retrospectively reviewed medical records of patients admitted to the ICU from January 2010 to May 2013. We included patients who were older than 15 years and had received mechanical ventilation (MV) for more than 48 hours. Patients at risk of imminent death on admission or treated by non-invasive positive pressure ventilation were excluded. We collected patient demographics, reasons for MV, arterial blood gas analyses and ventilator settings at 48 hours or later after beginning of MV. Performing multivariable logistic regression analysis to clarify independent factors related to hyperoxemia, we compared the hyperoxemia and non-hyperoxemia groups. Results: During the study period, 328 out of 1,664 patients were enrolled. Hyperoxemia occurred in 83 (25.3%) patients. Comparable values for each group were (hyperoxemia vs. non-hyperoxemia): PaO2, 133 (124–145) vs. 98 (84–108) mmHg; inspired O2 fraction, 0.3 (0.3–0.4) vs. 0.3 (0.3–0.4); positive end-expiratory pressure, 6 (6–8) vs. 8 (6–8) cmH2O; and duration of MV 144 (72–387) vs. 170 (105–325) hours. Hyperoxemia correlated with younger age (p = 0.027), lower serum lactate level (p = 0.029) and decompensated heart failure (p = 0.042). Multivariable logistic regression analysis revealed hyperoxemia to be independently associated with age ≤ 39 (odds ratio[OR], 2.6, 95% confidence interval [CI], 1.1 to 6.0) and decompensated heart failure (OR, 1.9, CI, 1.1 to 3.5). APACHE II score ≥ 30 was associated with lower incidence of hyperoxyemia (OR, 0.53, CI, 0.3 to 1.0). Conclusions: Hyperoxemia was common in younger and APACHE II score