Oxygen therapy and anaesthesia: too much of a good thing?

Oxygen is given to patients around the time of surgery to prevent or treat acute hypoxaemia, the harmful consequences of which are potentially augmented in the setting of the peri-operative inflammatory response. Whilst sub-acute and chronic hypoxaemia are frequently well tolerated by humans, both in health and illness 1, the adaptive responses to acute hypoxaemia are limited and intervention may be required to prevent harm. One of the many roles of the anaesthetist is to protect patients from significant hypoxaemia, and this commonly involves administering additional inspired oxygen alongside other interventions. That said, it does not necessarily follow that 'too much' oxygen is the best solution to 'not enough' 2; it is becoming increasingly clear that hyperoxaemia has the potential to be harmful in a variety of clinical scenarios 3. Although others have discussed the merits of considering a more conservative use of oxygen in medical practice 4 it is perhaps timely to re-evaluate anaesthetists' management of arterial oxygenation. Patients requiring general anaesthesia for surgery invariably receive supplemental inspired oxygen, both intra-operatively and for a variable period postoperatively. In addition to this, there are periods when high concentrations of inspired oxygen are administered as a precautionary measure to prevent unplanned catastrophic hypoxaemia: the 'oxygen bolus'. It is common practice to administer 100% inspired oxygen to patients at key points during the conduct of general anaesthesia, typically before induction and during emergence. The purpose of this pre-oxygenation is to replace nitrogen with oxygen within the lungs, primarily within the functional residual capacity (FRC), thus providing a reservoir of oxygen that can diffuse into the pulmonary circulation even if ventilation ceases. Whole body oxygen reserves can be increased from approximately 1500 ml to 4000 ml through this approach. However, it is important to remember that inspiring 100% oxygen has a minimal effect on arterial oxygen content (CaO2) because haemoglobin is already close to maximal saturation when breathing air and oxygen is poorly soluble in plasma. The adequacy of pre-oxygenation is best assessed by end-tidal oxygen fraction, and a target of 0.9 has been recommended 5. In healthy individuals, this simple intervention increases the time to desaturation (SpO2 < 90%) during apnoea from 1.0 to 6.9 minutes compared with breathing air 6. Desaturation below an SpO2 of 90% places a patient perilously close to the steep portion of the oxyhaemoglobin dissociation curve, where severe hypoxaemia may develop rapidly. One of the adverse consequences of such a pre-oxygenation strategy is pulmonary atelectasis. High concentrations of inspired oxygen result in absorption atelectasis, even after brief periods of therapy, and the magnitude of this effect is dependent upon the duration and concentration of oxygen administration 7, 8. Thus, a balance needs to be struck between the benefits of having a reserve (of oxygen and time) to minimise harm in case of an acute airway emergency, and the physiological harm of prolonged exposure to high concentrations of inspired oxygen, with the associated theoretical risk of increased postoperative pulmonary complications. It is always difficult to weigh rare but potentially catastrophic consequences (e.g. hypoxic brain damage) against common but incremental harms (e.g. hyperoxic pulmonary damage), particularly when both the likelihood and the severity of harm are dependent on individual susceptibility, around which there is additional uncertainty. Such dilemmas are the bread and butter of clinical decision-making, but in this context we have very limited data on which to base our judgements. Once the airway has been successfully secured, it becomes harder to justify the use of high concentrations of inspired oxygen, as the potential harm may start to outweigh any benefit. Whilst pre-oxygenation during induction may be appropriate in some patients, the universal use of 100% oxygen at the end of an operation when preparing for emergence from anaesthesia and extubation (or removal of a supraglottic airway device) is less clearly justified. High concentrations of inspired oxygen at the end of surgery increase the incidence of significant pulmonary atelectasis 9 and this can have significant clinical consequences 10. The recent Difficult Airway Society guidelines for the management of tracheal extubation recommend the use of 100% oxygen even in cases deemed to be at 'low risk' of airway incident. However, there is no discussion about, or reference to, the possible risks of hyperoxia in this document, and it is unclear whether such risks have been taken into account in its development 11. Again, we face the challenge of weighing rare catastrophic consequences against common incremental harm. Over and above the issues around intubation and extubation, there is also a general tendency for us, as anaesthetists, to reach for the oxygen rotameter during a wide range of intra-operative difficulties that are unrelated to hypoxaemia. This is the result of generations of teaching that if you can do nothing else, give oxygen. For potential emergencies involving the airway or respiratory system, high-concentration oxygen may have merit; for example a dislodged laryngeal mask airway or severe bronchospasm. But for a number of other acute intra-operative events, the use of oxygen may serve more to alleviate our own stress rather than providing any direct benefit to the patient. In haemorrhage, the deficit is red blood cells; delivering 100% inspired oxygen will not significantly improve convective oxygen carriage unless the patient was previously hypoxaemic. During cardiac ischaemia (ST depression or elevation), 100% inspired oxygen may cause intense coronary vasoconstriction and reduced coronary blood flow, thereby paradoxically lessening oxygen delivery to the myocardium. Following severe hypotension (possibly with a concomitant reduction in cardiac output), 100% inspired oxygen may cause a further reduction in stroke volume and cardiac output, primarily through an increase in systemic vascular resistance 12. Administration of 100% oxygen also decreases cerebral blood flow, which may not be a desirable response at such times 13. The use of 100% inspired oxygen to manage these intra-operative emergencies should be questioned, and may in future be reserved for those situations in which there is clear evidence of benefit or, at least, no suggestion of harm. There is also an often unrecognised (but clearly recorded) trend towards maintaining significantly higher than normal arterial oxygen partial pressure (PaO2), without adjustment of inspired oxygen fraction (FIO2), during major surgery. In an audit of 75 patients undergoing major elective surgery at one of our institutions, mean PaO2 on the first blood gas was 24.4 kPa, which did not change significantly throughout surgery (unpublished data). The rationale for delivering an FIO2 above 0.21 is based on a number of well-understood pharmacological and physiological sequelae of general anaesthesia that may lead to a reduction in arterial oxygenation. Drug-induced respiratory depression, a reduction in functional residual capacity (FRC), altered ventilation-perfusion matching, pain and partial airway occlusion all contribute to the likelihood that this alteration of normal physiology will occur. However, there is a high degree of inter-individual variability in this phenomenon, and in most cases it can usually be rectified by a modest increase in FIO2, to approximately 0.3 for most patients without significant cardiorespiratory co-morbidities. The maintenance of a PaO2 significantly higher than normal is interesting. Is this state of super-normal oxygenation maintained 'just in case' there is an unanticipated intra-operative crisis, or does this represent indifference to supra-normal oxygen values based on an assumption that there is no risk of harm? Paradoxically, keeping a patient at an artificially high PaO2 may actually mask any decline in respiratory function, due to the buffer created at the top end of the oxyhaemoglobin dissociation curve during hyperoxaemia. Whatever the reason, it seems worthy of reflection whether maintaining such a non-physiological milieu during a time of considerable tissue trauma and inflammatory stress responses is in the patient's best interests. The debate as to whether high-concentration oxygen (typically a FIO2 of 0.8) reduces postoperative surgical site infections has continued for some time. Each new study seems to swing the pendulum between benefit and no benefit. In February 2012, a meta-analysis of seven trials concluded that a high FIO2 was not beneficial for preventing surgical site infections 14, but six months later, a meta-analysis of nine trials reported benefit 15. Understandably, considerable clinical confusion exists in this area. Some trials were stopped prematurely 16, 17 and the statistical methodology of others reporting a beneficial effect of high FIO2 have been criticised 18. Worryingly, there are also data suggesting that high-concentration oxygen may be harmful. One study reported an incidence of surgical site infections of 25.0% (vs 11.3% in the control group) that led to a significantly longer hospital length of stay 19. Furthermore, in a post-hoc analysis of long-term follow-up of the PROXI trial (that showed no reduction in surgical site infections), increased long-term mortality was reported following a high FIO2 during abdominal surgery 20, 21, and cancer-free survival was significantly shorter in the high-inspired oxygen group 22. A similar debate exists over whether high-concentration oxygen is effective in reducing postoperative nausea and vomiting, something for which a biological mechanism seems less plausible than a reduction in surgical site infections. A meta-analysis of 10 trials in 2008 found there to be no benefit 23, whilst a more recent meta-analysis of 11 trials suggested that high inspired oxygen levels prevented the occurrence of late nausea 15. An important issue in this debate is whether there are plausible biochemical mechanisms to explain the clinical data suggesting harm from hyperoxia? A delicate balance exists in all cells between oxidation and innate anti-oxidant species. Imbalance favouring oxidation leads to oxidative stress, which in turn results in cellular injury, including impairment of mitochondrial function and damage to proteins and DNA through the excess formation of reactive oxygen species. Hyperoxia, inflammation and ischaemia-reperfusion all accelerate oxidation, whilst a deficit of antioxidant defences tips the balance further towards oxidative stress. Incomplete reduction of oxygen results in reactive oxygen species such as the superoxide and hydroxyl radicals and hydrogen peroxide. These molecules play a vital role in normal cellular signalling but, in excess, they can be highly destructive and have been implicated in a wide range of diseases, including cancer. The hydroxyl radical is one of the most reactive biological species ever discovered. The anti-oxidant systems exist to protect us from excessive radicals, but become overwhelmed in the face of prolonged hyperoxia. Other biological signalling molecules vital to normal homeostasis, such as nitric oxide, carbon monoxide and hydrogen sulphide, can also be affected by exogenous over-oxygenation of cells. Whilst oxidative stress may seem irrelevant to the average day-case procedure, it may not be for major high-risk surgery. Despite sophisticated risk stratification, it remains impossible to predict which patients will go on to suffer peri-operative complications that lead to critical illness. In this situation, hyperoxia-induced oxidative stress is a plausible mechanism of biological injury and may be linked to a worse outcome for some patients. Of particular relevance to this group is the risk of pulmonary oxygen toxicity, particularly in those who already have underlying lung pathology 24. Furthermore, there is a rational argument that during operations in which ischaemia-reperfusion injury is a major factor, such as organ transplantation, excessive tissue oxygen levels may exacerbate dysfunction in the transplanted organ, through up-regulated oxidative stress pathways 25, 26. A lack of adequate monitoring may be a factor in how we chose to manage peri-operative oxygenation. Pulse oximetry is a reasonable tool for the detection of hypoxaemia, but there is no evidence that using it alters the outcome of patients undergoing surgery 27, a fact that many anaesthetists find hard to rationalise. One possible explanation for this finding is that the absence of benefit from avoidance of hypoxaemia through the use of pulse oximetry may in part be due to unidentified harm from hyperoxaemia as a result of administering a FIO2 to ensure that normal oxygen saturations are maintained. Perhaps we should be more precise in our targeting of oxygen levels, avoiding both hypoxaemia and hyperoxaemia 1, 28. The implementation of novel devices to monitor tissue oxygenation more closely may be valuable in this respect. Near-patient continuous arterial blood gas monitoring, tissue perfusion monitors and servo-control mechanisms to regulate arterial oxygenation automatically may find a place in the future of anaesthesia for high-risk surgery and critical care medicine. The challenge of weighing up the risk-benefit equation for arterial oxygenation is that the immediate effects of acute hypoxaemia due to an anaesthetic mishap are devastatingly obvious, whereas the detrimental consequences of hyperoxaemia are difficult to measure in real time and have an impact that may only become apparent hours or days later. We emphatically do not advocate radical changes in practice today; such an approach may carry significant risk. However, we do suggest that thoughtful assessment of the risks and benefits for every patient in whom oxygen is administered is worthwhile. Oxygen is a drug, and like all drugs, precise dosage is important to achieve the optimal balance between benefit and harm. MG is funded in part by the British Oxygen Fellowship of the Royal College of Anaesthetists awarded by the National Institute of Academic Anaesthesia. MG is also executive chair of the Xtreme-Everest Oxygen Research Consortium. DM has received honoraria from Siemens Healthcare Diagnostics and is a Director of the Xtreme-Everest Oxygen Research Consortium. We would like to thank Dr Clare Morkane for the original data described in this editorial.