Oxygen : the elixir of life or the kiss of death?

The administration of supplemental oxygen is ubiquitous in medical practice, especially among critically ill patients. Hyperoxia is fairly common during oxygen therapy, and generally considered to be less deleterious than the potential harm that may arise from hypoxia. However, there has been an increased understanding of the detrimental effects of hyperoxia in recent times. How does the clinician balance the lifesaving effect of oxygen therapy while minimizing possible harm? 

Hypoxemia and compensatory mechanisms

A PaO2 level of less than 60 mm Hg is generally considered to be the lower limit of acceptable oxygenation. Several adaptive mechanisms are triggered in the presence of hypoxia. Downregulation of mitochondrial uncoupling has been demonstrated in human volunteers; hypoxia may enable more efficient ATP generation and mitochondrial protection.1 The release of hypoxia-inducible factors result in the activation of glycolytic enzymes leading to a preponderance of anerobic metabolism. Mitochondrial hibernation may occur, resulting in a reduction in the oxygen demand.2 Other compensatory mechanisms that enable acclimatization include hypoxic pulmonary vasoconstriction, increased cardiac output, polycythemia, and increased production of 2-3 diphosphoglycerate with a shift of the oxygen-dissociation curve to the right, enabling offloading of oxygen to the tissues.3

Potential harmful effects of hyperoxia

Hyperoxia results in the generation of reactive oxygen species (ROS). Excessive production of ROS resulting from high levels of oxygen may lead to cellular death by necrosis or apoptosis. Furthermore, ROS are the chief mediators of reperfusion injury. Hyperoxia has been well known to lead to pulmonary damage similar to acute respiratory distress syndrome, inhibition of mucociliary transport, and atelectasis. Supranormal arterial partial pressures of oxygen also lead to a fall in the cardiac output due to generalized vasoconstriction and increased afterload. Besides, coronary vasoconstriction may occur, and predispose to myocardial ischemia.4

Evidence of harm or lack of benefit from hyperoxia

In a randomized controlled study of normoxic patients with acute ST-elevation myocardial infarction (STEMI), supplemental oxygen at 8L/min was compared to no oxygen therapy. Recurrent myocardial infarction was significantly more common in the supplemental oxygen group; besides, there was an increase in the incidence of cardiac arrhythmias with oxygen supplementation. Infarct size assessed by cardiac magnetic resonance imaging (MRI) was significantly higher with oxygen therapy.5 In a similar study of normoxic patients with STEMI, supplemental oxygen was administered until the completion of percutaneous coronary intervention and compared to controls who received no supplemental oxygen. No difference was observed in the myocardial salvage index assessed by cardiac MRI or in the infarct size.6 These studies suggest that supplemental oxygen therapy in normoxic patients with acute myocardial infarction offers no benefit and may be potentially harmful. 

Does supplemental oxygen improve outcomes in acute stroke patients who are not hypoxic? Patients with acute stroke were randomized to receive continuous supplemental oxygen for 72 h, nocturnal oxygen only for three consecutive nights, or no oxygen in the Stroke Oxygen Study.7 No significant difference was observed in death or disability at 3 months, assessed using the modified Rankin score. 

The association between supranormal oxygen levels and in-hospital mortality was examined in a multicenter cohort study among survivors of cardiac arrest. A PaO2 level of > 300 mm Hg carried a significantly higher in-hospital mortality compared to 60–300 mm Hg or < 60 mm Hg.8 This study suggests that supranormal oxygen levels may worsen reperfusion injury following cardiac arrest. 

Three contemporaneous studies have compared a conservative to a more liberal oxygenation target in mechanically ventilated patients. The Oxygen ICU study randomized mechanically ventilated patients targeting a PaOlevel between 70–100 mm or oxygen saturation (SpO2) of 94–98% in the conservative group.9 In the liberal group, PaO2 levels of up to 150 mm Hg was allowed, with an oxygen saturation between 97–100%. ICU mortality was significantly lower in the conservative group (25% vs. 44%, p = 0.01). Panwar et al. studied patients who were likely to require mechanical ventilation for more than 24 h.10 In the conservative group, the target SpO2 was 88–92% compared to more than 96% in the liberal group. No significant differences were observed in new onset organ dysfunction, intensive care or 90-day mortality between groups.

An observational cohort study involving three Dutch ICUs assessed the association between predefined metrics of hypoxia and clinical outcomes.11 Mild hyperoxia was defined as a PaOranging between 120 to 200 mm of Hg and severe hyperoxia as a PaOmore than 200 mm Hg. Logistic regression analysis was performed with each metric of hyperoxia using specific thresholds. A significant association was observed between severe hyperoxia and mortality rates; ventilator-free days were also less in patients with hyperoxia. A linear relationship was seen between the duration of hyperoxia and hospital mortality. 

Permissive hypoxia 

A U-shaped relationship has been demonstrated between PaO2 and hospital mortality; the lowest mortality being between 110-150 mm Hg. Mortality was higher at a PaOof <67 mm Hg and >225 mm Hg.12 There has been increasing focus on “permissive” hypoxia, in patients who might suffer adverse outcomes from interventions aimed to maintain oxygenation in the physiological range. It is also important to consider lung injury resulting from high FiOlevels besides the harmful effects of systemic hypoxemia. Permissive hypoxia must ensure a favorable balance between oxygen consumption and delivery. The parameters that may help assess adequacy of oxygen delivery include lactate levels, central venous oxygen saturation (>70%), central venous to arterial PaCOdifference (<6 mm Hg), and the ratio between the central venous and arterial venous PaCO2difference and the arterial to central venous PaOdifference (<1.23).13 Clinical evidence to support permissive hypoxia is lacking; the threshold levels below which harm may ensue is unknown and likely to vary depending on age, the underlying clinical condition, and the chronicity of hypoxia. In practice, targeting lower values of PaO2 (55-80 mmHg) may be safer compared to the adverse outcomes that may ensue from efforts at maintaining normoxia. 

The bottom line

  • High levels of inspired oxygen and PaOhave been well established to cause harm. Aiming for physiological levels of oxygenation with injurious ventilatory strategies are likely to be counterproductive. 
  • Several damage-limiting compensatory mechanisms are triggered by hypoxia; however, hyperoxia may lead to unmitigated harmful effects. 
  • Supplemental oxygen therapy is not beneficial and may lead to adverse outcomes in normoxic patients with acute myocardial infarction and acute stroke. 
  • The is a growing body of evidence that supports the use of conservative targets for oxygenation in critically ill patients. 
  • Permissive hypoxia may be appropriate in selected patients; however, the safe thresholds for permissible oxygen levels are unknown. 


1.        Martin DS, Grocott MPW. Oxygen Therapy in Critical Illness: Precise Control of Arterial Oxygenation and Permissive Hypoxemia*. Crit Care Med. 2013;41(2):423-432. doi:10.1097/CCM.0b013e31826a44f6

2.         Schumacker PT. Hypoxia-inducible factor-1 (HIF-1). Crit Care Med. 2005;33(12 Suppl):S423-425.

3.         MacIntyre NR. Tissue Hypoxia: Implications for the Respiratory Clinician. Respir Care. 2014;59(10):1590-1596. doi:10.4187/respcare.03357

4.         Farquhar H, Weatherall M, Wijesinghe M, et al. Systematic review of studies of the effect of hyperoxia on coronary blood flow. Am Heart J. 2009;158(3):371-377. doi:10.1016/j.ahj.2009.05.037

5.         Stub D, Smith K, Bernard S, et al. Air Versus Oxygen in ST-Segment–Elevation Myocardial Infarction. :8.

6.         Khoshnood A, Carlsson M, Akbarzadeh M, et al. Effect of oxygen therapy on myocardial salvage in ST elevation myocardial infarction: the randomized SOCCER trial. Eur J Emerg Med Off J Eur Soc Emerg Med. 2018;25(2):78-84. doi:10.1097/MEJ.0000000000000431

7.         Roffe C, Nevatte T, Sim J, et al. Effect of Routine Low-Dose Oxygen Supplementation on Death and Disability in Adults With Acute Stroke: The Stroke Oxygen Study Randomized Clinical Trial. JAMA. 2017;318(12):1125. doi:10.1001/jama.2017.11463

8.         Kilgannon J. Hope, Jones Alan E., Parrillo Joseph E., et al. Relationship Between Supranormal Oxygen Tension and Outcome After Resuscitation From Cardiac Arrest.Circulation. 2011;123(23):2717-2722. doi:10.1161/CIRCULATIONAHA.110.001016

9.         Girardis M, Busani S, Damiani E, et al. Effect of Conservative vs Conventional Oxygen Therapy on Mortality Among Patients in an Intensive Care Unit: The Oxygen-ICU Randomized Clinical Trial. JAMA. 2016;316(15):1583. doi:10.1001/jama.2016.11993

10.       Panwar R, Hardie M, Bellomo R, et al. Conservative versus Liberal Oxygenation Targets for Mechanically Ventilated Patients. A Pilot Multicenter Randomized Controlled Trial. Am J Respir Crit Care Med. 2016;193(1):43-51. doi:10.1164/rccm.201505-1019OC

11.       Helmerhorst HJF, Arts DL, Schultz MJ, et al. Metrics of Arterial Hyperoxia and Associated Outcomes in Critical Care*: Crit Care Med. 2017;45(2):187-195. doi:10.1097/CCM.0000000000002084

12.       de Jonge E, Peelen L, Keijzers PJ, et al. Association between administered oxygen, arterial partial oxygen pressure and mortality in mechanically ventilated intensive care unit patients. Crit Care. 2008;12(6):R156. doi:10.1186/cc7150

13.       He H, Liu D, Long Y, Wang X. High central venous-to-arterial CO2 difference/arterial-central venous O2 difference ratio is associated with poor lactate clearance in septic patients after resuscitation. J Crit Care. 2016;31(1):76-81. doi:10.1016/j.jcrc.2015.10.017

Leave a Reply