The Pros and Cons of Oxygen in the ICU: Part 2

Aaron B. Holley, MD


April 01, 2016

Optimal PaO2 for ARDS

About a year ago, I commented on an article that suggested that a partial pressure of oxygen (PaO2) between 95 and 110 mm Hg was optimal for patients with acute respiratory distress syndrome (ARDS).[1,2] The argument was that driving up the PaO2 would optimize neurocognitive recovery, a laudable goal. I was skeptical, as were two reviews I cited as part of my discussion.[3] I concluded by stating that I'd need to see a randomized controlled trial (RCT) demonstrating that a higher PaO2 is beneficial. Well, an RCT evaluating oxygen targets in the intensive care unit (ICU) was recently published.

Investigating Whether Higher PaO2 Is Beneficial

Panwar and colleagues[4] enrolled 103 patients who were on mechanical ventilation (MV) for less than 24 hours and randomly assigned them to different oxygen saturation (SpO2), not PaO2, targets. Using a fraction of inspired oxygen (FiO2) between 0.21 and 0.80, the bedside clinicians targeted an SpO2 of 88%-92% in the conservative group and greater than 96% in the liberal group. The primary outcomes were all related to oxygen variables and not clinical endpoints. Secondary outcomes were clinical and included change in sequential organ function assessment (SOFA) score, new-onset ARDS, change in creatinine, incidence of hemodynamic instability, vasopressor-free days (to day 28), arrhythmia-free days (to day 28), ICU mortality, and 90-day mortality. Less than 30% of patients in the study had ARDS, and the reasons for ICU admission were otherwise heterogeneous and included medical (majority), surgical, and trauma patients.[5]

The authors found that a "permissive hypoxemia" approach is "feasible," but no secondary outcomes reached statistical significance. In fairness, the study wasn't powered to detect clinically significant differences. There were a few semi-interesting findings: (1) Patients in the liberal group spent more time on mandatory modes of MV; (2) 22% of the SpO2 readings in the liberal group were >96% (hyperoxia); (3) in a prespecified subgroup analysis (PaO2/FiO2 <300), the adjusted hazard ratio for mortality at day 90 in the conservative arm was 0.49 (95% confidence interval, 0.20-1.17; P=.10); and (4) vasopressor dosing was lower in the liberal arm. The authors speculated that clinicians felt that patients in the conservative group were eligible for weaning earlier, and this drove the decrease in use of mandatory modes of MV. This is difficult to prove, and because there was no difference in MV-free days between groups, it's of little consequence even if it's true.

More Evidence Is Needed

Unfortunately, this wasn't the RCT I was looking for when I concluded my earlier post. It's a great start, though, and the authors should be commended for completing a multisite RCT in which oxygen levels were successfully manipulated in a critically ill population. The authors of the RCT noted that there are two additional multicenter RCTs targeting oxygen levels in the ICU (NCT01319643 [OXYGEN-ICU] and NCT01722422 [HYPER-2S]) that were recently completed. According to, OXYGEN-ICU is designed to enroll 660 patients with 30-day mortality as the primary outcome. They're comparing normoxia (94%-98%) vs hyperoxia (>97%). HYPER-2S will be complete after enrollment of 441 patients. The trial design is more complicated than OXYGEN-ICU or the recently published Panwar paper. In addition to looking at oxygen levels, they're also assessing the effects of hypertonic saline. Again, the primary outcome is 30-day mortality.

In any case, it looks as if we'll be getting more information soon. The jury's still out on the appropriate oxygen target. As more randomized data are published, we'll start to obtain clarity. Until then, physiology would dictate that we continue to avoid hyperoxia, maintain permissive hypoxia when appropriate for patients with chronic obstructive pulmonary disease, and otherwise keep SpO2 above 90%.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.