Background
Early hyperoxia may be an independent risk factor for mortality in critically ill traumatic brain injury (TBI) patients, although current data are inconclusive. Accordingly, we conducted a retrospective cohort study to determine the association between systemic oxygenation and in-hospital mortality, in critically ill mechanically ventilated TBI patients.
Methods
Data were extracted from the Australian and New Zealand Intensive Care Society Centre for Outcome and Resource Evaluation Adult Patient Database. All adult TBI patients receiving mechanical ventilation in 129 intensive care units between 2000 and 2016 were included in analysis. The following data were extracted: demographics, illness severity scores, physiological and laboratory measurements, institutional characteristics, and vital status at discharge. In-hospital mortality was used as the primary study outcome. The primary exposure variable was the ‘worst’ partial arterial pressure of oxygen (PaO2) recorded during the first 24 h in ICU; hyperoxia was defined as > 299 mmHg. Adjustment for illness severity utilized multivariable logistic regression, the results of which are reported as the odds ratio (OR) 95% CI.
Results
Data concerning 24,148 ventilated TBI patients were extracted. By category of worst PaO2, crude in-hospital mortality ranged from 27.1% (PaO2 40–49 mmHg) to 13.3% (PaO2 140–159 mmHg). When adjusted for patient and institutional characteristics, the only PaO2 category associated with a significantly greater risk of death was < 40 mmHg [OR 1.52, 1.03–2.25]. A total of 3117 (12.9%) patients were hyperoxic during the first 24 h in ICU, with a crude in-hospital mortality rate of 17.8%. No association was evident in between hyperoxia and mortality in adjusted analysis [OR 0.97 (0.86–1.11)].
Conclusions
In this large multicenter cohort of TBI patients, hyperoxia in the first 24 h after ICU admission was not independently associated with greater in-hospital mortality. Hypoxia remains associated with greater in-hospital mortality risk and should be avoided where possible.
