(1565) A Sub-Analysis of Severe Traumatic Brain Injury Patients Enrolled in the SAVE-O2 Randomized Trial

Introduction: Hypoxemia is detrimental to critically ill patients, especially those with traumatic brain injury (TBI). However, a growing body of literature demonstrates that hyperoxemia (SpO2>96%) can also have deleterious outcomes after TBI. We performed a sub-analysis of the Strategy to Avoid Excess Oxygen (SAVE-O2) trial focusing on critically ill trauma patients with TBI.

Methods: SAVE-O2 was a multicenter cluster-randomized, stepped-wedge trial to target normoxemia in adult trauma patients admitted to the intensive care unit [NCT04534959]. We randomized eight academic hospitals to cross over from pre-implementation (usual care) to post-implementation (intervention) phases at three-month intervals. The intervention involved an educational and informatics program targeting normoxemia (SpO2 90-96%) and down-titrating supplemental oxygen in patients with hyperoxemia. In subjects identified subjects with TBI, we used descriptive, inferential, and multivariable methods to compare and adjust for cofounders. Our primary outcome was supplemental oxygen free days (SOFD), defined as the number of days alive and not receiving supplemental oxygen through day 28.

Results: The SAVE-O2 study enrolled 13,903 subjects across Level 1 trauma centers. Of those, 27% had a documented TBI. Overall, hyperoxemia time (per patient hour of exposure) decreased from 71% in the pre-intervention phase to 47% in the post-intervention phase in mechanically ventilated patients and from 22% to 13% in non-mechanically ventilated patients. Time spent in hypoxemia (SpO2 < 88%) was < 1% for all patients. Overall, the unadjusted SOFD increased from 19.3 days pre-implementation to 20.9 days post-implementation. After adjusting for covariates, the study intervention improved SOFD by +1.7 days (95%CI 1.1-2.4, p< 0.001). TBI patients had a greater improvement in SOFD (+1.8 days, 95%CI 0.8-2.8, p< 0.001) than non-TBI patients (+0.7 days, 95%CI -0.3-1.6, p=0.15) (p for interaction=0.02).

Conclusions: Among critically ill trauma patients, our preliminary analysis indicates the subgroup of patients with TBI may have experienced improved outcomes, compared to non-TBI patients, from an intervention targeting normoxemia.