Intended for healthcare professionals

Practice Lesson of the Week

Rebound hypoxaemia after administration of oxygen in an acute exacerbation of chronic obstructive pulmonary disease

BMJ 2011; 342 doi: (Published 31 March 2011) Cite this as: BMJ 2011;342:d1557
  1. Binita Kane, SpR in Respiratory Medicine1,
  2. Peter M Turkington, consultant physician in respiratory medicine1,
  3. Luke S Howard, consultant physician and honorary senior lecturer in respiratory medicine2,
  4. Anthony G Davison, consultant physician in respiratory and acute medicine3,
  5. G John Gibson, emeritus professor of respiratory medicine4,
  6. B Ronan O’Driscoll, consultant physician in respiratory medicine1
  1. 1Manchester Academic Health Science Centre, University of Manchester, Salford Royal Foundation Trust, Salford M6 8HD
  2. 2Hammersmith Hospital, Imperial College Healthcare NHS Trust, London, UK
  3. 3Southend University Hospital, Westcliff-on-Sea, UK
  4. 4University of Newcastle upon Tyne, Newcastle upon Tyne, UK
  1. Correspondence to: Dr B R O’Driscoll, Department of Respiratory Medicine, Salford Royal University Hospital, Stott Lane, Salford M6 8HD, UK ronan.o’driscoll{at}

Abrupt removal of high concentration oxygen in chronic obstructive pulmonary disease may produce life threatening hypoxaemia

Acutely ill patients with chronic obstructive pulmonary disease may be given high concentration oxygen in ambulances and emergency departments despite evidence that this is potentially harmful.1 Supplementary oxygen can correct life threatening hypoxaemia. However, acutely ill patients with chronic obstructive pulmonary disease given uncontrolled oxygen therapy may develop worsening hypercapnic (type II) respiratory failure, potentially leading to severe respiratory acidosis and coma.2 3 4 Abrupt removal of oxygen can provoke hypoxaemia that is more severe than before oxygen treatment was started.3 5 We illustrate this phenomenon with the following case report.

Case report

An 86 year old woman with a history of chronic obstructive pulmonary disease was admitted to hospital with increasing breathlessness. She was a former smoker (40 pack years of smoking). She had hypertension and a history of myocardial infarction one year previously with moderate left ventricular impairment on an echocardiogram. She used an Accuhaler inhaler (GlaxoSmithKline) delivering 500 µg fluticasone, 50 µg salmeterol twice daily, a tiotropium inhaler once daily, and a salbutamol inhaler as needed. She was severely disabled by breathlessness but had no previous admission to hospital for lung disease. Arterial blood gases 15 months previously (figure, column A) showed moderate hypoxaemia but normal arterial carbon dioxide pressure. When she was stable after recovery from the current episode, arterial blood gases showed that she had developed chronic type II respiratory failure (figure, column F).


Fig 1 Arterial blood gas measurements in our patient (see table 1 in the appendix on for values of arterial carbon dioxide pressure and arterial oxygen pressure. The table also shows normal ranges according to the Salford Royal Foundation Trust’s laboratory)

Her chest radiograph showed minor consolidation at both lung bases with mild pulmonary oedema and a small left sided pleural effusion. The white blood cell count and C reactive protein were raised. She was treated for an infective exacerbation of chronic obstructive pulmonary disease and heart failure with nebulised bronchodilators, antibiotics, steroids, furosemide, ramipril, and oxygen. Her arterial oxygen saturation recorded by pulse oximetry (SpO2) while she was breathing 24% oxygen was 96%. Arterial blood gases were not measured initially, but the admitting doctor documented that oxygen therapy should be given at “24-28% via a Venturi mask, aiming for a target SpO2 range of 88-92%.”

Twenty four hours after her admission she developed a fever and became unwell, and her oxygen saturation (pulse oximetry) had fallen to 85%. A nurse increased the oxygen to 4 litres per minute via a nasal cannula. A doctor who was called to review the patient documented drowsiness and imminent danger of respiratory arrest. Her oxygen saturation (pulse oximetry) was 100%. A sample of arterial blood gases was taken (figure, column B). The doctor concluded that she had acute on chronic type II respiratory failure exacerbated by excessive oxygen therapy and removed the oxygen, with a plan to refer the patient for non-invasive ventilation if the blood gases did not improve. The figure (column C) shows the results of her arterial blood gases while breathing room air 30 minutes later. The arterial carbon dioxide pressure had fallen only slightly, but the arterial oxygen pressure and arterial oxygen saturation had fallen dramatically, to 4.2 kPa and 59.6% respectively. The patient was then treated with oxygen at 1 litre per minute via a nasal cannula, with improvement of the blood gases after several hours (figure, column D). Further measurements two hours later while the patient was breathing room air (figure, column E) again showed severe hypoxaemia and persistently high arterial carbon dioxide pressure. She subsequently recovered fully and was discharged home.


Administration of high concentrations of oxygen to patients with chronic obstructive pulmonary disease can lead to increasing hypercapnia because of complex mechanisms, which include worsening ventilation-perfusion mismatching and reduced alveolar ventilation.3 In our patient, this resulted in clinical deterioration and impaired consciousness. Supplementary oxygen therapy was discontinued, and the patient breathed room air until the arterial blood gases were repeated 30 minutes later (figure, C). The arterial oxygen pressure had fallen dramatically from 19.8 to 4.2 kPa, whereas the arterial carbon dioxide pressure had fallen by only 1.5 kPa, and she remained severely hypercapnic. The reason for the disparity between the reductions in arterial oxygen pressure and arterial carbon dioxide pressure is that the body’s oxygen stores are limited, whereas stores of carbon dioxide are much larger owing to its high solubility in the tissues, extracellular fluid, and blood.6 Consequently, stopping oxygen therapy results in a more rapid decline in arterial oxygen pressure than in arterial carbon dioxide pressure. Importantly, when oxygen treatment has led to an increase in arterial carbon dioxide pressure, its abrupt removal causes the arterial oxygen pressure to fall to a level lower than that present before oxygen therapy was started.

This “rebound hypoxaemia” can be explained as follows: in the alveoli, carbon dioxide and oxygen compete with each other for space; if alveolar carbon dioxide pressure rises to a level that is higher than that present before oxygen treatment was started and if oxygen is then stopped abruptly, the alveolar oxygen pressure will fall to a level lower than before oxygen therapy was started, with a consequent fall in arterial oxygen pressure. Death caused by severe hypoxaemia may result. (See the appendix on for a detailed explanation using the alveolar gas equation.)

Although our patient survived, we know of other patients who have died suddenly when oxygen therapy was stopped abruptly in the presence of severe hypercapnia, and in those cases acute arterial hypoxaemia, rather than worsening of hypercapnic respiratory acidosis, was probably the cause of death.

Patients with severe hypercapnia and acidosis secondary to excessive oxygen should be considered for immediate ventilatory support. If uncontrolled oxygen has led to dangerous hypercapnia, oxygen should not be discontinued, but the delivery system should be changed to one that allows accurate control of inspired oxygen concentration and withdrawal should be gradual rather than abrupt. Depending on the severity of hypercapnia, a compromise is to change to a Venturi mask delivering 28% or 35% oxygen. This should ensure an adequate alveolar oxygen pressure, sufficient to protect the patient from severe rebound hypoxaemia while awaiting the start of ventilatory support.3 For example, application of the alveolar gas equation shows that when the arterial carbon dioxide pressure rose to 12 kPa on uncontrolled oxygen, a 28% mask would have produced an alveolar oxygen pressure of at least 11.6 kPa. If the arterial carbon dioxide pressure had risen as high as 15 kPa, a 35% mask should ensure an alveolar oxygen pressure of at least 14.6 kPa (an adequate pressure).

The key clinical message for junior doctors and others in this situation is that while waiting for expert help, oxygen should not be removed abruptly owing to the risk of severe and potentially fatal rebound hypoxaemia. Controlled oxygen should be administered via 28% or 35% Venturi mask until the patient can be assessed for ventilatory support.


Cite this as: BMJ 2011;342:d1557


  • Contributors: BRO’D had the idea for the article. BK performed the literature search and wrote the original article. BRO’D, GJG, LSH, AGD, and PMT were responsible for revisions to the article before submission. BRO’D is the guarantor.

  • Funding: None.

  • Competing interests: All authors have completed the Unified Competing Interest form at (available on request from the corresponding author) and declare: no support from any organisation for the submitted work; no financial relationships with any organisations that might have an interest in the submitted work in the previous three years; and no other relationships or activities that could appear to have influenced the submitted work.

  • Provenance and peer review: Not commissioned; externally peer reviewed.

  • Patient consent obtained.


View Abstract