Is Hyperoxemia in Trauma Bad?
/AIMS
Severe trauma is the leading cause of death worldwide for adults younger than 50 years of age. Acute traumatic life support (ATLS) guidelines endorse early and aggressive usage of supplemental oxygen in patients with severe trauma, at least until abnormalities of airway or breathing can be safely ruled out. However, unclear target concentration, duration or saturation goals often leads to hyperoxemia. Emerging studies in the intensive care unit (ICU) setting suggest that liberal supplemental oxygen therapy and hyperoxemia is associated with increased mortality. Limited evidence in the trauma population suggests similar outcomes. Simultaneously, the suggestions recommended by ATLS are not founded on strong evidence and instead represent expert opinion or consensus. Further evidence with animal models suggests that hyperoxemia may be detrimental even in the first 8 to 12 hours of oxygen administration. Thus, the authors of the TRAUMOX2 trial aimed to compare restrictive vs liberal supplemental oxygen in the early phase of resuscitation for acute trauma patients with regards to mortality and major respiratory complications within 30 days of injury.
METHODS AND STATISTICS
The trial investigators enrolled and randomized 1979 patients across five trauma referral centers in Denmark, Netherlands and Switzerland. Patients were eligible to be enrolled if they were older than 18 (or appeared child-bearing age) with blunt or penetrating trauma if the injury was severe enough to suggest a hospital stay longer than 24 hours.
Patients were excluded if cardiac arrest occurred or the providers suspected carbon monoxide poisoning.
Notably, this was a superiority trial with the hypothesis that targeting normoxia would show favorable outcomes as compared to a liberal oxygen supplementation strategy. Based on the TRAUMOX1 trial, the authors assumed a 33% relative risk reduction in the primary outcome in the intervention group compared to the control group.
Enrollment and randomization were completed in a 1:1 ratio with block enrollment stratified based on site of inclusion (prehospital vs trauma center) and presence of endotracheal intubation at the time of randomization.
The trial was designed to target simple, pragmatic monitoring and interventions and therefore data was primarily collected utilizing finger SpO2 monitors with periodic analysis utilizing arterial blood gases at 1 and 6 hours after randomization. Because providers could not be blinded to SpO2 values, the trial was considered “open-label.”
INTERVENTION
Patients were randomized to either the restrictive or liberal oxygenation arm. Those in the restrictive oxygen group received the lowest dosage of oxygen possible to maintain an SpO2 value of 94% either utilizing no supplemental oxygen, a nasal cannula, nonrebreather mask or mechanical ventilation. If the patient achieved an O2 saturation >94% on room air, they were allowed to remain in the restrictive group.
Patients in the liberal oxygen group received 15 liters/minute (L/min) of oxygen via nonrebreather mask for non-intubated patients and a fraction of inspired oxygen (FiO2) of 100% in those on a mechanical ventilator. This could subsequently be reduced to 12 L/min or an FiO2 of 60% as long as the SpO2 remained greater than or equal to 98%.
Both groups received targeted oxygen therapy for eight hours after enrollment. All other care, traumatic or otherwise, was provided by standard set by ATLS and local institutions.
RESULTS
The primary outcome was a composite of death or major respiratory complications within 30 days of randomization. Major respiratory complications included pneumonia (as based on US Centers for Disease Control criteria) or acute respiratory distress syndrome (based on the Berline definition). This was assessed by blinded examiners at each site.
Interestingly several “exploratory” outcomes were measured as well including episodes of hypoxemia, ICU readmission, sepsis, surgical site infection, and pneumonia post-discharge. These were used as hypothesis-generating endpoints for future studies.
The primary outcome occurred in 118/733 (16.1%) patients in the restrictive oxygen group and 121/724 (16.7%) patients in the liberal oxygen group with an odds ratio of 1.01. Pre-defined subgroups and adjustments made with per-protocol analysis showed similar results. Across all groups, there was no significant difference in the primary outcome. When considered individually, both death and major respiratory complications were statistically unchanged between groups. Interestingly, the trends suggested major respiratory complications might be less likely in the restrictive group while death rates sightly favored the liberal group.
With regards to exploratory outcomes, the restrictive oxygen group showed a statistically significant decrease in surgical site infections as compared to the liberal oxygen group. Unsurprisingly, hypoxemic episodes were statistically less likely in the liberal oxygen group. No differences were noted in the rates of ICU readmission, sepsis or pneumonia after discharge.
LIMITATIONS
Several limitations were discussed by the authors. First, they note that no data was collected on race and ethnicity which may limit the interpretation of pulse oximetry in relation to skin pigmentation. While they note that an 8-hour intervention period might be too brief to impact clinical outcomes, they lay out support evidence that hyperoxemia leads to rapid generation of reactive oxygen species and leads to definitive harm in animal models.
Notably, the average time to randomization across groups was 30 minutes, with greater than one quarter of patients being enrolled beyond the 50-minute mark. The patients may have received any degree of supplemental oxygenation prior to enrollment ranging from nothing to full hyperoxygenation. If hyperoxygenation is indeed harmful so rapidly, this contamination period of nearly 1-hour may have impacted the study greatly.
As in any open-label study, the non-blinded providers may have (consciously or unconsciously) changed nonoxygen interventions due to the patient’s randomization in the study. Another source of statistical bias comes from the post-randomization exclusion of 471 patients. This occurred due to patients being enrolled only to have minor (or no) injuries that led to the patient being discharged. The bias arises from the fact that patients placed on supplemental oxygen (that is, the liberal group) would be less likely to be discharge regardless of traumatic injuries and therefore the exclusions could be skewed toward the restrictive group.
SUMMARY
This trial revealed that in a large, multinational European population with both blunt and penetrating trauma, restrictive or liberal oxygen supplementation in the initial 8-hour period of resuscitation did not significantly alter outcomes for patients with regards to mortality or respiratory complications.
Interestingly, those with restrictive oxygenation may suffer from fewer surgical site infections, although future trials will need to be designed to more adequately answer this question.
Two sub-study analyses have been published utilizing the TRAUMOX2 cohort. The first showed that rates of hypoxemia lasting greater than 5 minutes were relatively rare and did not differ significantly between restrictive or liberal oxygen supplementation. Additionally, there were no significant differences in biomarkers of oxidative stress between the groups.
Authorship
Written by Cody Stothers, PGY-3, University of Cincinnati Dept of Emergency Medicine
Audio Editing by Anita Goel, MD, Assistant Professor, University of Cincinnati Dept of Emergency Medicine
Editing and posting by Jeffery Hill, MD MEd, Associate Professor, University of Cincinnati Dept of Emergency Medicine
Cite as: Stothers, C. Hill, J., Goel, A. Is Hyperoxemia is Trauma Bad? TamingtheSRU. www.tamingthesru.com/blog/journal-club/hyperoxemia. 5/7/2025.