Oxygen therapy in acute medical care

BMJ 2002; 324 doi: https://doi.org/10.1136/bmj.324.7351.1406 (Published 15 June 2002) Cite this as: BMJ 2002;324:1406
  1. Alastair J Thomson (a.j.thomson{at}ed.ac.uk), research fellow,
  2. David J Webb (d.j.webb{at}ed.ac.uk), Christison professor of therapeutics and clinical pharmacology,
  3. Simon R J Maxwell (s.maxwell{at}ed.ac.uk), senior lecturer,
  4. Ian S Grant (i.s.grant{at}ed.ac.uk), consultant anaesthetist
  1. Clinical Pharmacology Unit, Department of Medical Sciences, Western General Hospital, University of Edinburgh, Edinburgh EH4 2XU
  2. Intensive Care Unit, Western General Hospital

    The potential dangers of hyperoxia need to be recognised

    Oxygen is cheap, widely available, and used in a range of settings and conditions to relieve or prevent tissue hypoxia. Since its discovery by Scheele and Priestley in the 1770s, it has remained one of the most effective therapeutic agents available. However, as a result of poor prescribing and monitoring, inappropriate doses are often given.1

    Oxygen is most commonly delivered by devices with variable performance such as face masks and nasal cannulae. These can produce unexpectedly high concentrations of inspired oxygen, particularly when ventilation is depressed.2 In addition, masks that incorporate a reservoir bag are often used in emergencies, following the widespread adoption of advanced trauma life support style (or ATLS) guidelines. These appliances can produce systemic hyperoxia that is generally assumed to be harmless. However, emerging evidence suggests that for some patients with acute medical conditions, hyperoxia may be harmful.

    Oxygen therapy is often provided for patients with acute coronary syndromes, although the evidence to support its benefits is controversial. Animal studies have shown that hyperoxia can reduce coronary blood flow,3 particularly to ischaemic areas.4 Hyperoxia can also produce potentially adverse effects on systemic haemodynamics—reducing cardiac output and increasing blood pressure and systemic vascular resistance—in patients with myocardial infarction or congestive heart failure. 5 6 These haemodynamic changes may further impair blood flow and delivery of oxygen to tissues already hypoperfused. This apparent paradox occurs because supraphysiological oxygen tensions produce little increase in the oxygen carrying capacity of blood, but they promote reflex vasoconstriction via local regulatory mechanisms in arteriolar smooth muscle.7 In one of the few randomised double blind controlled trials of oxygen therapy, patients with uncomplicated myocardial infarction randomised to receive oxygen tended to have a higher mortality and more ventricular tachycardia than those randomised to receive air.8 Routinely giving oxygen to patients with a stroke has also been questioned. Indeed, a recent randomised study showed an increase in mortality at one year in non-hypoxic patients with mild or moderate strokes who had received supplemental oxygen as part of their initial treatment.9

    The dangers of pulmonary toxicity due to oxygen, ranging from atelectasis to pro-inflammatory processes, and the risks of prescribing oxygen for patients with chronic obstructive pulmonary disease are more widely appreciated. A recent study investigated the prevalence of respiratory acidosis in patients with an exacerbation of chronic obstructive pulmonary disease. This confirmed that injudicious use of oxygen therapy is associated with respiratory acidosis, which then predicts a subsequent requirement for intensive care.10 This study emphasises that treatment guidelines should be based on achieving target arterial oxygen tensions and saturations rather than on giving predetermined concentrations or flow rates of inspired oxygen.

    Inevitably, oxygen continues to be given on the basis of presumed need rather than of documented low oxygen saturation. This may be acceptable practice in emergencies. In more controlled circumstances, however, efforts should be made to confirm the presence of hypoxia by pulse oximetry before giving oxygen. In this way normoxic patients, who do not require treatment with oxygen, will avoid the risks associated with hyperoxia.

    Pulse oximeters should be available in all clinical areas where acutely ill patients are managed and oxygen is given, allowing oxygen saturation to be monitored on a continuous basis. At the very least, oxyhaemoglobin saturation should be documented before and intermittently during oxygen therapy. Supplemental oxygen can then be given on a rational basis focusing on the attainment of target oxygen saturations, which may differ between patient groups, rather than on using fixed flow rates. However, patients requiring high inspired oxygen concentrations to maintain adequate oxygen saturations must always be cared for in areas where intensive monitoring is available. In addition, arterial blood gas analysis can be done if necessary to confirm that appropriate oxygen and carbon dioxide tensions, and pH, have been achieved. In this way patients can get oxygen therapy that is tailored to their need and the unfavourable cardiopulmonary effects of hyperoxia can be minimised. The use of oxygen prescription charts similar to those advocated by Dodd and colleagues may, in non-specialist areas, help achieve these aims.1

    Oxygen therapy remains a cornerstone of modern medical practice. To further quantify the risks associated with hyperoxia more trials are needed. The results of well conducted trials may lead to refinements in the use of oxygen. Unfortunately, due to its accepted role in therapeutic practice and virtually non-existent potential for commercial development, oxygen therapy has attracted little research funding in recent years. At present doctors should strive to ensure that oxygen is prescribed, administered, and monitored with care. This will enable us to achieve optimal tissue oxygenation for more of our patients.


    1. 1.
    2. 2.
    3. 3.
    4. 4.
    5. 5.
    6. 6.
    7. 7.
    8. 8.
    9. 9.
    10. 10.
    View Abstract