By Samuel Nadler, MD, PhD

Critical Care, Pulmonary Medicine, The Polyclinic, Madison Center, Seattle; Clinical Instructor, University of Washington, Seattle

Dr. Nadler reports no financial relationships relevant to this field of study.

SYNOPSIS: Patients randomized to maintain oxygen saturation between 94-98% experienced better outcomes than patients allowed to receive partial pressure of oxygen > 150 mmHg.

SOURCE: Girardis M, et al. Effect of conservative vs. conventional oxygen therapy on mortality among patients in an intensive care unit. JAMA 2016;316:1583-1589.

Supplemental oxygen is a ubiquitous therapy in the ICU, although precise targets for oxygen saturation (SpO2) and arterial partial pressure of oxygen (PaO2) are unclear. Although it is clear that hypoxia can lead to harm, data showing that hyperoxia produces harmful effects are more limited. The OXYGEN-ICU trial was a single-center, randomized, clinical study that examined the effects of oxygen delivery on mortality in patients in the ICU. Adult patients with an expected stay in the ICU of > 72 hours were randomized to either conservative or conventional oxygen therapy. Patients in the conservative group received supplemental oxygen titrated to maintain SpO2 between 94-98% or PaO2 between 70-100 mmHg. Patients in the conventional control group received supplemental oxygen according to standard ICU protocols targeting SpO2 97-100% and allowing PaO2 values up to 150 mmHg. Patients who were pregnant, immunosuppressed, or transitioned to comfort measures only were excluded. This study was meant to enroll 660 patients for adequate power to detect a 6% mortality difference but was terminated early due to an earthquake, limiting further enrollment and an unscheduled interim analysis demonstrating significant benefit with the conservative protocol. Overall, this study enrolled 480 patients, with 236 randomized to conservative oxygen therapy and 244 to conventional therapy. Although the two groups were randomized, some important differences in each group are notable. Fewer patients in the conservative arm presented with COPD, chronic liver disease, respiratory failure, mechanical ventilation, shock, liver failure, renal failure, and documented infections overall, leading to a lower Simplified Acute Physiology Score II (SAPS II) in the conservative group compared to the conventional arm (37 vs. 39, respectively). Trial conductors used a modified intention-to-treat model was used that censored patients enrolled in the trial who left the ICU within 72 hours. Thus, 216 patients were included in the conservative analysis and 218 in the conventional group. With these caveats, conservative therapy led to an absolute risk reduction in ICU mortality of 8.6% (11.6% vs. 20.3% in the conservative vs. conventional groups, respectively; P = 0.01). With conservative oxygen therapy, there also were improvements in many secondary outcomes, including hospital mortality, shock, bacteremia, and mechanical ventilation-free hours. A subgroup post-hoc analysis of only patients on mechanical ventilation also demonstrated mortality benefit (absolute risk reduction 5%; 95% confidence interval, 0-9%).


At first, this study seems to indicate that targeting more modest SpO2 goals of 94-98% may be beneficial to ICU patients. However, several caveats deserve mention. First, this study was terminated early due to factors beyond the control of the investigators, but early termination predisposes to overestimation of effect size. The study was not terminated due to pre-specified futility criteria as might affect other trials terminated due to safety concerns. While randomized, the conservative therapy arm clearly contained lower-acuity patients, and this could certainly influence the outcomes. Overall, there were few events in each arm, which also can affect the reliability of these data. However, this study shows trends similar to other studies of conservative vs. conventional oxygen delivery. The AVOID investigators demonstrated that supplemental oxygen administered to patients with acute ST-segment elevation myocardial infarction without oxygen desaturation led to larger infarct size and may increase early myocardial injury.1 The PROXI trial demonstrated that higher FiO2 during anesthesia for abdominal surgery led to an increased likelihood of infection.2 In contrast, a smaller pilot study from the CLOSE and ANZIC investigators failed to show significant effects on mortality in mechanically ventilated patients.3

This study adds to the body of evidence that supplemental oxygen delivery without regard to need can be harmful. More importantly, this study seems to indicate that there is no strong indication to supplement SpO2 levels close to 100%, and that patients with SpO2 of 94% may not need supplemental oxygen in the ICU. Targeting more modest SpO2 and PaO2 goals clearly is not harmful and may lead to certain benefits.


  1. Stub D, et al. Air versus oxygen in ST-segment elevation myocardial infarction. Circulation 2015;131:2143-2150.
  2. Meyhoff CS, et al. Effect of high perioperative oxygen fraction on surgical site infection and pulmonary complications after abdominal surgery. JAMA 2009;302:1543-1550.
  3. Panwar R, et al. Conservative versus liberal oxygenation targets for mechanically ventilated patients. Am J Respir Crit Care Med 2016;193:43-51.