Hyperoxia in the ICU: Is less more?
With the growing body of evidence, the need of the hour was an ICU-based randomized trial that may settle the debate. The 21 center, 1,000 patient ICU-ROX trial promised to deliver on that (Mackle D, et al. N Engl J Med. 2019 Oct 14. doi: 10.1056/NEJMoa1903297). The study design was more reflective of real-life clinical scenarios than some of its predecessors, with the control group exposed to usual-oxygen therapy instead of liberal hyperoxia. Both groups had a lower saturation threshold of 91% while the conservative-oxygen group had an upper limit of 97% along with a conscious effort made to drop the FIO2 to 21%. Though both groups had similar median PaO2 levels, the conservative group spent much greater time (median 29 hours) at 21% FIO2 than the usual group (median 1 hour). SpO2 targets also allowed frequent changes to oxygen delivery without the need for blood gases.
Presuming the primary effect of oxygen toxicity would be on the lungs, the study was powered for a primary outcome of ventilator-free-days, which showed no significant difference among the groups. No significant differences in mortality or other secondary outcomes were observed.
The ICU-ROX trial leaves us with a few questions, the most important are:
Are the detrimental effects of hyperoxia limited to certain disease-specific groups or generally applicable?
The evidence is substantial inpatients with cardiac arrest/myocardial injury. A prespecified subgroup analysis in ICU-ROX indicated a higher number of ventilator-free days with conservative oxygen therapy in patients with hypoxic ischemic encephalopathy. When asked, Dr. Paul Young, one of the investigators of the ICU-ROX group, states, “These are actually pretty small subgroups, and the number of mortality events is quite small. My belief is that these data are best viewed as hypothesis generating rather than practice changing”
Where do we stand?
While we look for further answers regarding the consequences of hyperoxia, it is established that conservative oxygen therapy aimed at reducing delivered FIO2 is a safe practice without any adverse outcomes. The conservative oxygen group in ICU-ROX allowed SpO2 levels as low as 91% with no serious hypoxic events. On the other hand, the IOTA group in their data analysis suggested a possible increase in mortality risk, which was dose-dependent on the magnitude of increase in SpO2, in the range of 94% to 96%. Based on the available evidence, it is reasonable to encourage targeting lowest FIO2 values needed to maintain SpO2 between 91% and 96% in our ICU patients. There would always be a small fraction of patients, such as those with ARDS or severe hypoxic respiratory failure, in whom this may not be achievable given fluctuating and unreliable SpO2 levels in the setting of profound hypoxia.
What lies ahead?
As the debate rages on, in an effort to answer this question for once and for all, the researchers of ICU-ROX are planning to conduct a multinational, multicenter RCT, the MEGA-ROX. An ICU trial of this size has not been attempted before and, given the sample size, Dr. Young feels the MEGA-ROX will be powered to detect an absolute mortality difference as low as 1.5%, if it does exist. There is a distinct possibility that conservative oxygen therapy will be best for patients with some diagnoses while liberal oxygen will be best for patients with other diagnoses. “We are conducting a number of parallel nested trials within the overall 40,000 participant trial sample. Each of these nested trials will evaluate a prespecified hypothesis in a specific cohort of critically ill patients and is accompanied by an appropriate power calculation. This will be able to address any heterogeneity of treatment effect among the different subgroups,” he concluded. As we eagerly await the results of MEGA-ROX, there may be a growing belief among intensivists that when it comes to oxygen in the ICU, less may be truly more.
Dr. Chaaban and Dr. Sen are with the University of Kentucky College of Medicine, Lexington, Kentucky.
Correction, 4/10/20: An earlier version of this article misstated Dr. Sen's name
