When used appropriately, oxygen can save lives and is one crucial element in the resuscitation of the critically ill patient and in peri-operative management. It is the most commonly used drug in emergency medicine and one of the most commonly used drugs in hospitals; 34% of ambulance patients are treated with oxygen and an estimated 15–17% of UK hospital inpatients are receiving oxygen at any one time [1, 2]. However, there are concerns that the use of oxygen can be haphazard and poorly documented. For example, only 32% of UK hospital patients who were receiving supplementary oxygen in 2008 had a valid prescription or any other written documentation of oxygen use . This is of concern because there is increasing evidence that inappropriate use of oxygen can be detrimental, resulting in significant morbidity and mortality. Despite these concerns, guidelines for acute oxygen use were not developed until 2008 .
Supplemental oxygen is employed in a number of clinical situations. It is essential in the correction of potentially harmful hypoxaemia, but increasing the fraction of inspired oxygen (FIO2) is only one element in the resuscitation process. Consideration of the cause of hypoxaemia is needed – addressing airway patency, circulatory competency, adequacy of cardiac output and the potential presence of respiratory depressants. Oxygen may reduce dyspnoea in the severely hypoxaemic individual but there is no evidence to suggest that supplemental oxygen reduces dyspnoea in the normoxaemic or mildly hypoxaemic patient [4, 5]. When used inappropriately, supplemental oxygen can, at best be useless and at worst, have significant detrimental effects. There are two main clinical concerns with the administration of supplemental oxygen: the potential development of hypercapnia; and the harmful effects of hyperoxaemia itself.
The danger of excess oxygen and its association with hypercapnia is well recognised in patients with chronic obstructive pulmonary disease (COPD). Mortality at the time of acute exacerbation is increased with the use of a higher FIO2 [6, 7]. There are approximately 100 000 COPD admissions per annum in the UK, with an associated mortality of 7.5% [6, 8]. The use of a controlled and titrated lower FIO2 may reduce mortality to 4% or possibly as low as 2% . Therefore, there is potential to prevent a few thousand deaths from COPD each year in the UK alone. There is also recent evidence that excessive oxygen therapy can cause hypercapnia in acute asthma as well as in COPD . This is an important observation because hypercapnia is a marker of near-fatal asthma and patients with this complication may require intubation and ventilation .
The risk of developing hypercapnia must not be underestimated. When audited, around 22% of arterial blood gases taken in our hospital showed evidence of hypercapnia, and in addition, almost three quarters of patients with documented risk factors for the development of hypercapnia had oxygen saturations above 92%, the suggested safe upper limit for these patients [3, 12]. This risk exists not only in COPD and severe asthma, but also in other conditions such as cystic fibrosis, bronchiectasis, chest wall disorders, neuromuscular disease and obesity hypoventilation .
Hyperoxaemia as a result of excess oxygen administration has been reported to have detrimental effects in a number of clinical settings. In ischaemic heart disease, hyperoxaemia causes coronary vasoconstriction, which reduces overall myocardial oxygen delivery and may also increase the risk of re-perfusion injury after myocardial infarction, in turn having the potential to extend infarct size and increase mortality [13, 14]. In the peri-operative setting, high concentrations of inspired oxygen can cause a reduction in functional residual capacity by producing areas of under-ventilation with resultant atelectasis and intra-pulmonary shunting, and may increase the incidence of acute lung injury [15-17].
An association with increased mortality in the first 24 hours of intensive care has been reported in over-oxygenated patients when compared with normoxaemia and, to a lesser degree, hypoxaemia . Similar findings have been reported in survivors of cardiac arrest . However, other authors have reported that most of the excess mortality amongst hyperoxaemic patients receiving critical care (including survivors of cardiac arrest) was abolished after adjusting for potential confounding factors in multivariant analysis, hypothesising that hyperoxia due to delivery of an increased FIO2 may be a marker of severity of illness, rather than have a causal link to mortality [20, 21]. Hyperoxaemia may also be a risk factor for increased mortality in patients with mild or moderate stroke .
In the past, supplementary oxygen was given routinely in the immediate postoperative period but, for the majority of patients, it is likely to be of no clinical value. The previously held beliefs that administration of high levels of oxygen (e.g. 80%) would enhance patient safety and could relieve nausea and vomiting have been questioned and now largely disproven [15, 23]. The clinical role of hyperoxia for preventing surgical site infection remains uncertain because randomised controlled trials on this topic have reported disparate results . The latest meta-analysis of this subject had negative overall outcomes but reported that there may be benefit for some subgroups such as patients undergoing colorectal surgery, and further studies were recommended .
As the oxygen dissociation curve is sigmoid-shaped, increases in partial pressure of oxygen will result in minimal increases in oxygen saturation once the upper plateau has been reached at about 98%. This will have little effect on oxygen carrying capacity given the latter's dependence predominantly on oxygen saturation and not partial pressure. Moreover, inappropriate use of supplemental oxygen in this setting reduces the potential for pulse oximetry to detect clinically important hypoventilation and could delay recognition of postoperative deterioration . It was hoped that oximetry in the peri-operative period would enable early recognition and correction of hypoxaemic events, which in turn could reduce morbidity and mortality and possibly reduce admissions to, and lengths of stay on, intensive care units. A recent Cochrane review identified five randomised controlled trials (22,992 patients) examining the use of peri-operative oximetry . The use of oximetry did lead to increased detection of hypoxaemia and resulted in changes in patient care – either by increasing the FIO2 given or prompting the use of naloxone. When oximetry readings were made available to clinicians, hypoxaemia was detected early and hence was 1.5-3 times less common (compared with patients whose oximetry results were concealed from the clinicians). However, despite early recognition and correction of hypoxaemia, there was no difference in complications, postoperative cognitive function, length of inpatient stay or mortality. This finding was at odds with the subjective opinion of many of the anaesthetists who took part in the studies. When questioned, 18% of anaesthetists reported situations in which they felt that the use of oximetry had prevented the occurrence of a potentially serious adverse event, but the objective evidence did not support these clinical impressions. The authors suggested that correcting modest hypoxaemia, by increasing the blood oxygen saturation from marginal to satisfactory, may confer no benefit to patients; however, only a small number of studies were available for review and further work is required for clarification in this area .
In this complex clinical context, the British Thoracic Society (BTS) Guideline that was produced in 2008 attempted to address many of the issues surrounding the use and prescribing of oxygen. This guideline, which is endorsed by the Royal College of Anaesthetists, set out a number of key principles and made recommendations regarding target saturation ranges, based on a combination of oxygen levels that were believed to be safe and those that were normal or near-normal. The central principle of the guideline is that oxygen is a treatment for hypoxaemia and clinicians should aim to achieve oxygen saturation levels that are appropriate for the patient's condition. The Guideline also states that oxygen is not useful for the treatment of dyspnoea in the absence of hypoxaemia, or for increasing organ perfusion in a normoxaemic patient .
The normal oxygen saturation range for a healthy young adult is 96–98%, diminishing slightly with advances in age [27, 28]. Specialist publications have suggested a safe lower limit of 90% for arterial blood oxygen saturation in critical illness, and about 88% in sepsis [29, 30]. However, these recommendations originate from critical care areas with intensive patient monitoring, rather than general hospital wards, and the BTS currently recommends that in the majority of patients, the target saturation range should be 94–98% – a compromise between what is normal and what is safe . In patients at risk of hypercapnia, the target range was set lower, at 88–92%. This is based on a number of observations, gleaned primarily from patients with COPD. Firstly, in acute exacerbations of COPD, saturations greater than 85% (equivalent to a PaO2 of about 6.7 kPa) prevent death . Secondly, nearly half of patients with exacerbations of COPD have hypercapnia and the majority of hypercapnic patients with a PaO2 in excess of 10 kPa (equivalent to a saturation of about 92–93%) have associated acidosis . Hence, a target range of 88–92% is recommended in patients at risk of hypercapnia . The lower limit is set at 88% rather than 85% to allow for inaccuracies of up to 3% in pulse oximeter readings.
A report by the UK's National Patient Safety Agency in 2010 identified at least nine deaths (and potentially up to 35 deaths) between 2004 and 2009 that were attributable to incorrect oxygen therapy, including four cases of insufficient oxygen therapy and four cases of excessive oxygen therapy . Equipment failure such as empty or disconnected oxygen supplies, or accidental connection to air outlets instead of oxygen outlets, accounted for most of the incidents associated with under-use of oxygen. It is likely that these figures are gross under-estimates because over-use of oxygen in COPD alone may cause a few thousand avoidable deaths in the UK each year, as estimated above, and there may be large numbers of additional deaths associated with hyperoxaemia amongst survivors of cardiac arrests and critically ill patients including those with stroke. Additionally, there are risks due to the combustion-enhancing properties of oxygen: there are documented instances of fires in homes and ambulances caused by the use of oxygen near sources of ignition, and a letter in this edition of Anaesthesia demonstrates in graphic detail that the same risks are ever-present in hospitals . MacDonald has reviewed the risk of fire and explosions caused by oxygen and other gases used during and after anaesthesia .
In summary, if used appropriately, oxygen can be a great friend. Treating potentially harmful hypoxaemia, it is a key element in successful resuscitation for many critically ill patients. However, inappropriate use of oxygen in the non-hypoxaemic patient, or failure to appreciate the risk of hypercapnia and adjust target saturation ranges accordingly, can turn oxygen into a deadly foe, needlessly increasing mortality and morbidity in a wide range of conditions. Consensus guidelines (summarised in Table 1) recommend a target saturation range of 94–98% for most hospital patients, with a lower target range of 88–92% for patients at risk of hypercapnia . The guidelines also recommend that oxygen should always be prescribed (except in emergences, when it should be documented and prescribed once the patient's condition has stabilised) and there should be clear documentation of oxygen administration (device and flow rate) at the bedside. Adherence to these guidelines should achieve optimal blood oxygen levels for all hospital patients and thus reduce the serious adverse effects that are associated with inappropriate under- or over-use of oxygen. Finally, the overriding message ‘handle with care’ is particularly apposite whenever oxygen is administered.
|1. Critical illness requiring high levels of supplemental oxygen|| |
Give 15 l.min−1 via a reservoir mask and once stable, reduce oxygen to aim for SpO2 of 94-98%.
If patient at risk of hypercapnic respiratory failure, aim for the same initial saturation as all other critically ill groups pending ABG.
|2. Serious illness requiring moderate amounts of oxygen if the patient is hypoxaemic|| |
Initially give 2–6 l.min−1 via nasal cannulae or 5–10 l.min−1 via facemask, aiming for SpO2 of 94–98%. If SpO2 can't be maintained or initial SpO2 < 85%, use a reservoir mask with 15 l.min−1.
If at risk of hypercapnic respiratory failure, aim for SpO2 of 88–92%, adjusting to 94–98% if the ABGs show a normal PaCO2.a Repeat ABG after 30–60 min.
|3. COPD/other conditions at risk of hypercapnic respiratory failure requiring low dose/controlled oxygenb|| |
Before ABG use a 28% Venturi mask (4 l.min−1), aiming for SpO2 of 88–92%, adjusting to 94–98% if the ABGs show a normal PaCO2.a Repeat ABG after 30–60 min.
If patient has an oxygen alert card, aim for the target range which is specific to him/her.
If patient is hypercapnic and acidotic despite 30 min of appropriate treatment and oxygenation, consider non-invasive ventilation.
|4. Conditions for which the patient should be closely monitored but oxygen is not required unless hypoxaemic||As for no. 2, above.|