Journal Logo


InFocus: Think Twice Before Routinely Administering Oxygen

Roberts, James R. MD

Author Information
doi: 10.1097/01.EEM.0000437839.37882.dd
    Figure. E
    Figure. E:
    mergency physicians like to have the monitor show the pulse oximetry at 100% so supplemental oxygen is a routine adjunct for many ED patients even when they are not hypoxic, and its administration is often continued after admission for no good reason. Oxygen is a toxin, however, that can actually increase mortality when used in high doses over prolonged times, even for critically ill ventilated patients. Even patients with acute myocardial infarction do not need routine high-dose supplemental oxygen, and its use does not decrease complications. Keep the pulse oximetry 96% or above with low-flow nasal cannulas in the ED.

    Noncritical patients in the ED frequently receive supplemental oxygenation, even when their pulse oximetry is totally normal. It is not uncommon to apply the oxygen nasal cannula/mask to patients with MIs, abdominal pain, or altered mental status, and it's often continued throughout the ED and for most of the hospital stay.

    Few physicians believe that oxygen toxicity is an issue with routine supplementation and especially believe high-flow oxygen is beneficial to the cyanotic newborn, the patient with a fresh myocardial infarction, and all critically ill patients on a ventilator. This is hardly the case. The prolonged use of oxygen can cause problems, and it's clearly not always beneficial to critically ill patients.

    Significance of Arterial Hyperoxia and Relationship with Case Fatality in Traumatic Brain Injury: A Multicentre Cohort Study

    Rincon F, Kang J, et al

    J Neurol Neurosurg Psychiatry

    2013 Jun 21. [Epub ahead of print]

    The authors of this complex retrospective multicenter study aimed to determine whether hyperoxia was associated with a higher incidence of in-hospital fatalities in ventilated traumatic brain-injured patients in the ICU. The study cohort consisted of 1,212 ventilated traumatic brain-injured patients treated at 61 U.S. hospitals between 2003 and 2008. Hyperoxia was defined as PaO2 greater than 300 mm Hg, and hypoxia was defined as PaO2 less than 60 mm Hg. The primary outcome was defined as in-hospital death.

    The problems associated with long-term disability in traumatic brain injury are gargantuan and well known. Multiple strategies are used to identify and optimize physiological derangements during early resuscitation, presumably because they might be important to limit secondary brain injury or to improve long-term outcome. Managing cerebral perfusion and organ blood flow is one of the important issues. It was always thought that cerebral ischemia was a major contributor to secondary brain injury, but recent studies have demonstrated a nonintuitive and different concept.

    Oxygen is necessary to maintain brain metabolism after injury, but it is now thought that excessive oxygen can actually potentiate brain injury by exaggerating the production of oxygen free radicals, triggering cellular injury and apoptotic cascade, and organ specific hypoperfusion. Numerous studies have demonstrated the detrimental effects of excessive hyperoxia in all forms of brain injury and in other critically ill patients. The authors of this study attempted to evaluate observational data related to this problem, determined the occurrence of hyperoxia upon admission to an ICU, studied the impact of hyperoxia on the inpatient case fatality rate, and attempted to determine whether the hyperoxia upon admission was an early indicator of in-hospital death after other confounders were evaluated. Obviously, this was a difficult issue to interpret.

    Patients were classified as demonstrating normoxia, hyperoxia, or hypoxia with a primary outcome measure being in-hospital fatality. Thirty-three percent of the admitted patients were normoxic, 46 percent were hypoxic, and 21 percent were hyperoxic. A variety of cardiovascular, metabolic, respiratory, renal, and neurological functions were monitored for all groups. Overall, 33 percent (400/1212) met the primary outcome of an in-hospital death, so these were sick patients with significant brain injuries. Fatality was highest in the hyperoxic group (41%), followed by the hyperoxic group (33%) and the normoxic group (23%).

    It is unclear whether the more seriously injured patients received less or more oxygenation. The authors concluded that exposure to hyperoxia was prevalent and associated with a lower likelihood of survival after hospital admission of brain-injured individuals. They contend that this was true even after controlling confounding variables. They also determined that exposure to hyperoxia was an independent predictor of in-hospital fatality. One mechanism was a decrease in cerebral blood flow secondary to the induced hyperoxia, which can decrease cerebral blood flow up to 33 percent. Other studies have demonstrated the detrimental effects of hyperoxia, which may include deterioration of cardiac output and vasoconstricting effects.

    The authors conclude that ventilated TBI patients should not be subjected to arterial hyperoxia, but should be treated instead with methods to produce normoxia. Per their conclusions, unnecessary high-flow oxygen delivery should be avoided in critically ill ventilated TBI patients.

    Comment: These authors conclude that unnecessary oxygen delivery increases TBI patient fatalities. Hyperoxia, defined as the PaO2 greater than 300 mm Hg, was deemed ultimately harmful. It is likely that most emergency physicians, when faced with a brain-injured patient requiring intubation and ventilation, would opt for the initial delivery of 100% oxygen, with the possibly mistaken belief that it would be optional for brain function.

    Patients staying in the ED would likely remain on high-flow oxygen until they were admitted to the ICU, where this mistake may be reiterated. These authors note that a PaO2 less than 300 mm Hg is desirable. This revelation is similar to prior but since-debunked recommendations that brain-injured patients should be hyperventilated to decrease the pCO2 to decrease cerebral edema. It is now concluded that hyperventilation should be avoided, as should hypoxia and hypercarbia. A pCO2 of 30-35 mm Hg is the goal in brain-injured ventilated patients.

    Association between Arterial Hyperoxia Following Resuscitation from Cardiac Arrest and In-hospital Mortality

    Kilgannon JH, Jones AE, et al



    The authors from Cooper University Hospital in Camden, NJ, note that supplemental oxygen is often administered in high concentrations to patients after cardiac arrest, but this universal intervention has recently come under scrutiny as being potentially harmful. It is generally agreed that too little oxygen can potentiate any oxygen injury, but too much oxygen may increase free radical production and trigger cellular death.

    These authors sought to determine whether exposure to hyperoxia after the spontaneous return of circulation from a cardiac arrest was associated with a poor clinical outcome. They studied patients who survived cardiac arrest to ICU admission, and attempted to determine whether the presence of post-resuscitation hyperoxia, defined as PaO2 greater than 300 mm Hg, was a common occurrence and whether it was ultimately associated with lower survival or possible discharge.

    This multicenter study of more than 6,000 patients found hyperoxia in 18 percent, hypoxia in 63 percent, and normoxia in 19 percent. The hyperoxia group manifested a significantly higher in-hospital mortality rate compared with the normoxic group (63% versus 45% respectively). The authors concluded that cardiac arrest patients admitted to the ICU following resuscitation had higher in-hospital mortality when hyperoxia was an independent variable. These authors also found that post-resuscitation hyperoxia was common when blood gas analysis was performed after ICU arrival. They attempted to control for a predefined set of confounding variables in a multivariable analysis, and when this was accomplished, they found that exposure to hyperoxia was an independent predictor of in-hospital death.

    Hyperoxia was also associated with a lower likelihood of independent functional status at hospital discharge compared with normoxia. A poor clinical outcome associated with hyperoxia was an unexpected finding, and the authors urge caution in interpreting their data. Because reperfusion after an ischemic insult results in a surge of reactive oxygen species, many of which may overwhelm natural antioxidants defenses, the oxidative stress from hyperoxia formed after reperfusion is thought to lead to increased cellular death by diminishing mitochondrial activity, disrupting normal enzyme activity, and damaging lipid membranes through peroxidation.

    When the pulse oximetry reads 100%, the PaO2 can be anywhere from 80 to 400 mm Hg, so an arterial blood gas must be used to ferret out the potential partial pressure produced by the inspired oxygen. Once the patient is stabilized, a PO2 of 94-96% is the goal, even with traumatic brain injury or while the patient is on a ventilator.

    It was noted that the American Heart Association still recommends 100% oxygen administration during resuscitative efforts. Many physicians frequently maintained the high FiO2 for variable lengths of time, however, after circulation had been successfully restored. Nearly one in five patients had exposure to hyperoxia post-cardiac arrest, and almost half of those patients had a PaO2 greater than 400 mm Hg. Post-resuscitative arterial hyperoxia appeared to be a common occurrence. Recent AHA recommendations have suggested targeting an arterial oxygen saturation not to exceed 94% to 96%. This study did not evaluate whether therapeutic hypothermia was attempted.

    Comment: The harm from long-term high-flow oxygen use is a relatively new concept, but it is grossly underestimated by most clinicians. It is not known to be detrimental during a short ED stay, but prolonged ED boarding and continuation of ED protocols makes this an important issue for emergency medicine. I would note that no study has found that hyperoxia has long-term beneficial effects in patients with any illness in the hospital. These two articles suggest that a cardiac arrest and traumatic brain injury are adversely affected by prolonged periods of hyperoxia, and that this situation should be avoided. Of course, it requires an arterial blood gas to determine the actual PO2, and a pulse oximeter of 100% can have a tremendously high arterial saturation or a normal oxygen saturation, yet another reason to perform an ABG in the ED after things have settled down.

    Investigators could find no significant difference in mortality, arrhythmias, the use of analgesics, or other cardiac parameters in post-myocardial infarction patients who were administered oxygen as far back as 1976. They could find no evidence that routine administration of oxygen in complicated myocardial infarction was beneficial. (Br Med J 1976;1[6018]:1121.) Oxygen administration causes an increase in systemic vascular resistance and a vasoconstrictive effect. It has long been known that cerebral, renal, and retinal vasoconstriction, as well as decreased coronary blood flow, occurs during inhalation of high concentrations of oxygen. Merely producing an elevated PaO2 can cause a toxic milieu for a variety of organs. These authors concluded that routine oxygen administration to all patients with myocardial infarction has little place unless hypoxia was obvious.

    Click and Connect! Access the links in EMN by reading this issue on our website or in our iPad app, both available


    Current Oxygen Administration Guidelines from the American Heart Association

    The 2010 guidelines from the American Heart Association list potential adverse effects of oxygen administration during adult and neonatal resuscitation. They recommend 100% oxygen during initial resuscitation from cardiac arrest, but providers should then titrate oxygen to the lowest level required to achieve an arterial oxygen saturation of 94%. This is thought to reduce oxygen toxicity while not being harmful to post-resuscitative care. An arterial oxygen saturation of 100% may correspond to a PaO2 anywhere between 80 and 500 mm Hg, so it is always appropriate to wean the FiO2 when the saturation is 100% and aim to keep the saturation maintained at greater than 94%. This was a class 1 recommendation by the AHA in 2010.

    The guidelines note that normal infants generally do not reach extrauterine values of blood oxygen levels until approximately 10 minutes after birth; that is important to keep in mind when resuscitating newborns. Hemoglobin saturation may normally remain at 70-80% for several minutes following a normal birth, and even produce the appearance of cyanosis during that time, but the clinical assessment of skin color is a poor indicator of oxyhemoglobin saturation in the neonatal period.

    The AHA said insufficient or excessive oxygenation can be harmful to the newborn. Even brief exposure to excessive oxygen during a normal delivery may be harmful. No studies have compared the outcome of neonatal resuscitation initiated with oxygen concentration or a targeted oxyhemoglobin saturation, but the American Heart Association believes that initiating resuscitation with room air or a blended oxygen mixture is better for the neonate than 100% oxygen. The use of 100% oxygen to resuscitate even a cyanotic baby is withheld until after 90 seconds of bradycardia and cyanosis.

    It appears that it's normal for an infant to remain blue for 30-50 seconds following normal delivery. The first breath does not give a newborn an oxygen concentration high enough to reverse the cyanosis that has been present in utero. It might very difficult for most physicians to withhold 100% oxygen in every delivery, but the current thinking is that inspired room air will do just fine, and high oxygen concentrations should be avoided unless the infant is not progressing normally.

    Reader Feedback: Readers are invited to ask specific questions and offer personal experiences, comments, or observations on InFocus topics. Literature references are appreciated. Pertinent responses will be published in a future issue. Please send comments to

    © 2013 by Lippincott Williams & Wilkins