Low vs. high oxygenation targets in trauma patients: which is better?
What does TRAUMOX2 bring to the debate?
Trauma patients are routinely provided with high doses of supplemental oxygen—even when they are relatively young, relatively healthy patients without lung disease with normal oxygen saturations and PaO2.
The origins of this practice are unknown, and no good evidence supports it; the theory seems to be that a larger oxygen reservoir could somehow provide reserves to regions of localized tissue hypoxemia, or to a normoxemic patient who abruptly deteriorates.
Guidelines for the initial care of trauma patients advise supplemental oxygen be provided even when oxygen levels are normal (ATLS), and that FIO2 80% be provided in the perioperative period (WHO).
Over time, as the potentially delirious effects of hyperoxia became better understood, some in the trauma community began to question the practice.
Most clinicians answering a survey disagreed with the guidelines, at least in Denmark.
A single-center trial that was stopped early for low enrollment found a massive 8.6% absolute mortality reduction with a restrictive strategy. Systematic reviews showed that hyperoxia might be harmful in trauma patients. A Cochrane meta-analysis limited to ICU patients found the evidence to be too poor quality to make any conclusions between a restrictive and a liberal strategy.
There seems to have been no suggestion that targeting normoxemia was dangerous, but guidelines remained unchanged and hyperoxia has continued as a common practice.
A randomized trial was conducted to help settle the question. Did it?
TRAUMOX 2 Trial
At five trauma centers and 15 pre-hospital bases in Europe in 2022-2023, 1,979 patients with severe trauma were randomized to receive either a restrictive oxygen strategy, targeting an arterial oxygen saturation of 94%, or liberal oxygen (providing around 15 L per minute of oxygen by nasal cannula or 100% FiO2 if intubated, titrating down as low as 60% FiO2 if oxygen saturation remained at or above 98%).
This was specified in the protocol for the first 8 hours of care, and then patients could be treated to any oxygen target.
The outcome was a composite of death and severe respiratory complications, which were defined as pneumonia by the CDC’s definition, or severe hypoxemic respiratory failure with pulmonary infiltrates, i.e. acute respiratory distress syndrome or ARDS.
Care teams were aware of patients’ allocations, but blinded critical care physicians adjudicated the respiratory outcomes after viewing charts with the oxygen delivery redacted.
The sample size was chosen based on event rates in the pilot study, TRAUMOX1 (which enrolled only 41 patients), with statistical cushion added in case event rates were lower than those seen in the pilot.
Results
1,508 patients completed the trial; 471 were excluded after randomization, mostly because they were found not to have severe trauma after their initial stabilization.
There was no significant difference in the primary composite outcome of death or major respiratory complications within 30 days: 16.1% in the restrictive oxygen group and 16.7% in the liberal oxygen group, with an odds ratio of 1.01 [95% CI, 0.75 to 1.37; P = .94].
Interestingly though, the two components of the primary outcome diverged:
liberal oxygen brought a numerically lower risk of death, 7.3% versus 8.6% in the restrictive group.
Major respiratory complications were numerically fewer in the restrictive group, 8.9% versus 10.8%.
Neither was statistically significant.
Atelectasis occurred less often in the restrictive group compared to the liberal oxygen group 27.6% versus 34.7%.
Limitations
It may seem implausible to expect to discover a 33% relative reduction in death or pneumonia from a reduced oxygen dose for 8 hours.
The rationale came from this retrospective cohort study of 211 patients and follow-up from the PROXI randomized trial of 1,400 patients, which found that delivering 80% FiO2 instead of 30% during a single abdominal surgery resulted in no clear short-term harm, but a 5% absolute increase in mortality two years later! (The mortality was concentrated in patients with cancer.)
Although TRAUMOX2 had a large drop-out rate after randomization, conducting randomized trials in trauma patients is very challenging. Patients had to be enrolled quickly and some error rate was unavoidable.
Conclusions
The opposite directions of the primary outcome components—death and pneumonia/ARDS—suggest that TRAUMOX2’s findings are noise. Or the effects were very small, and the trial was underpowered to detect them.
Noise may be all that can be found. Mounting evidence suggests the degree of oxygenation does not significantly influence outcomes in trauma or other critical illness, although no trial has been definitive.
In this sense, TRAUMOX2 was weakly reassuring toward that null hypothesis.
For more statistically robust answers, all eyes are on Mega-ROX, “a series of multicenter, multinational, randomized, parallel-group, registry-embedded clinical trials nested within an overall 40,000-patient trial sample.” To answer the question of heterogeneity of response, patients with many different illnesses are enrolled (trauma, brain injuries, etc). Mega-ROX is reportedly more than halfway to target enrollment and might provide results in the next few years.
It would be a really strange dose/response effect if a 4% increase in the oxygenation of hemoglobin really did result in a 33% reduction in death.
There’s only so much hemoglobin. Loading it up 98% vs 94% only gets a small amount more oxygen to the tissues. If your goal is to deliver more oxygen, seems to me that you need to pay more attention to the amount of hemoglobin.