Titrating Oxygen During and After Cardiopulmonary Resuscitation

Abstract

Listen

Translate article

SEVERAL INVESTIGATIONS HAVE REKINDLED IMPORtant concern that administration of 100% oxygen during and early after resuscitation from experimental cardiopulmonary arrest might be deleterious to the brain. In a canine model of ventricular fibrillation cardiopulmonary arrest, use of 100% oxygen compared with use of room air early during resuscitation was associated with increased neuronal death in selectively vulnerable brain regions and worse neurological outcome. Several studies have focused on oxidative injury to key mitochondrial enzymes (such as pyruvate dehydrogenase or manganese superoxide dismutase) or mitochondrial lipids (such as cardiolipin) in mediating these deleterious effects. Concern about the use of 100% oxygen in resuscitation is not new. In neonatal resuscitation, detrimental effects of 100% oxygen have been described in case reports and randomized controlled trials. However, infants have compromised antioxidant defenses and age-related differences in endogenous defenses against hypoxemia, including fetal hemoglobin, among others. Thus, the potential risk of 100% oxygen and the potential benefit of room air may be greatly magnified in neonates compared with adults. In this issue of JAMA, Kilgannon et al report the results of an important multicenter cohort study generated from a critical care database of intensive care units (ICUs) in 120 US hospitals. The authors studied 6326 adults with nontraumatic cardiopulmonary arrest and analyzed the relationship between in-hospital mortality and hypoxia (PaO2 60 mm Hg), normoxia (PaO2 300 mm Hg), or hyperoxia (PaO2 300 mm Hg) as assessed on the first ICU arterial blood gas. Hyperoxia was associated with a significantly increased mortality rate compared with normoxia (proportion difference, 18%; 95% confidence interval [CI], 14%-22%). Moreover, the hyperoxia group showed increased mortality vs the hypoxia group (proportion difference, 6%; 95% CI, 3%-9%). In a model controlling for a predefined set of confounders, hyperoxia exposure had an odds ratio for death of 1.8 (95% CI, 1.5-2.2). In this study, 18% of the patients had hyperoxia based on the first arterial blood gas determination in the ICU. Given the rather conservative definition of hyperoxia (PaO2 300 mm Hg), the true incidence of more moderate levels of hyperoxia is likely to be quite high. Even though mechanisms producing secondary deleterious effects after cardiac arrest can be successfully manipulated (as evidenced by the use of induction of mild hypothermia), this finding underscores the possibility that further meaningful improvements in outcome might result from careful attention to appropriately titrating basic aspects of extracerebral physiology at the bedside, such as prevention of hyperoxia. The authors acknowledge the limitations of this observational study. For instance, it would have been informative to have provided an assessment of the temporal relationship of hyperoxia with outcomes because experimental work suggests the possibility that early hyperoxia rather than delayed postresuscitation hyperoxia is deleterious to the brain. The first PaO2 value in this study was obtained within 24 hours of ICU arrival, precluding assessment of the temporal effects of hyperoxia. Moreover, cause of death and neurological outcome were both lacking in the database, limiting any inferences regarding the contribution of cerebral vs extracerebral effects to the reported findings. The authors suggest putative deleterious effects of hyperoxia on pulmonary function, but (based on experimental data) tissue hyperoxia early postreperfusion also could adversely affect other organ systems. This study also showed an association between hypoxia and mortality after cardiopulmonary arrest. Many underlying pathologies that may require high levels of fraction of inspired oxygen to achieve normal arterial saturation can either cause or represent comorbidities in adults with cardiopulmonary arrest, such as drowning or pulmonary embolism. This complicates the ability to make sweeping recommendations against the use of 100% oxygen early in resuscitation. Similarly, it is not clear that arterial hyperoxia necessarily results in brain tissue hyperoxia. As