Screen Shot 2015-08-14 at 13.26.45

“Knee-Jerk Medicine” – Oxygen Therapy, COPD and Critical Illness



Owing to a widespread misconception affecting everyone from medical students to senior medical practitioners, patients with COPD are often deprived of life-saving oxygen therapy during times of critical illness.


Quite rightly we are taught at medical school to be aware of the risk of precipitating type II respiratory failure by giving uncontrolled oxygen therapy in patients with COPD. This is owing to a reliance on ‘hypoxic drive’ to maintain respiratory rate and therefore ventilation. So, if we give too much oxygen we obliterate the hypoxia and the patient’s central nervous system no longer feels the need to drive the respiratory musculature.


However, there is a crucial caveat that is poorly understood. This theoretical risk of ‘obliterating hypoxic drive’ is not relevant to all patients with COPD. In fact it may not be relevant to most patients with COPD. According to BTS guidelines those patients most dependent on hypoxic drive are those with resting hypoxia (i.e. on home oxygen), severe COPD (i.e. a high degree of airflow limitation on spirometry, poor exercise tolerance, and clinically severe disease suggested by frequent exacerbations), or a history of NIV use.


Therefore it is only these patients with severe disease that we should be wary of giving uncontrolled oxygen therapy to. It is not good practice to deny all COPD patients of precious oxygen therapy.


Let me discuss some cases to illustrate this.




Whilst working as the on-call medical registrar I was asked by my SHO to review a breathless patient with oxygen saturations of 80% on air. I advised the SHO over the phone to apply high-flow oxygen as I made my way over, which she did (but with disapproval from the nursing staff – again owing to this widespread misconception).


I arrived to find a 60 year-old gentleman severely tachypnoeic with a respiratory rate of 42 breaths per minute, saturations of 92% on high-flow oxygen and widespread crepitations on his chest. There was reasonably good chest wall excursion on inspection (i.e. he was shifting air pretty well) and no wheeze heard (i.e. no significant bronchospasm or airflow limitation). He was haemodynamically stable.


I diagnosed acute pulmonary oedema and treated him appropriately with IV furosemide and sublingual GTN (nursing staff weren’t happy to put up GTN infusion). Serial arterial blood gases (on high-flow oxygen, remember) revealed a pO2 of 9 kPa and a pCO2 of 6.6 kPa with a pH of 7.2. There was also an underlying metabolic acidosis – bicarbonate was 18 and base excess was minus 5.


The arterial gas remained the same on two gases performed 40 minutes apart with the patient on high-flow oxygen in the intervening period. Furthermore, this patient had an oxygen saturation of 95% on air prior to this episode of pulmonary oedema.


In other words, this gentleman was peri-arrest with profound hypoxaemia and whilst you would classify his ABG as showing type II respiratory failure, the pCO2 was only slightly raised (not high enough to account for a pH of 7.2) and it was not climbing. To all intents and purposes, both clinically and biochemically, this patient was suffering from a pulmonary diffusion problem (due to his lungs being water-logged) not an airflow obstruction problem.


Unfortunately, because ‘probable COPD’ had been written in the notes all of the medical staff were panicking about using high-flow oxygen, and looking at me in disbelief when I said he doesn’t need nebulisers (consider that he doesn’t have a wheeze, and also that nebulisers cannot be delivered with high-flow oxygen – so we would be giving him an unnecessary treatment and depriving him of life-saving oxygen).


Whilst he may have underlying COPD (life-long smoker) that is not his main problem now. I asked him if he had been bothered with a cough, wheeze or phlegm production in the preceding 6 months – he said no. Considering that COPD is characterized by chronic cough, wheeze and phlegm there certainly isn’t strong evidence for diagnosing severe COPD clinically here.


With the patient not improving I called the ITU registrar for a review. He kindly attended and reviewed the patient. In anticipation of a knee-jerk “sats 88-92%” response I took pains to carefully document that this gentleman had a stable pCO2 on high-flow oxygen, remained significantly tachypnoeic (i.e. not developing respiratory depression), and that his clinical picture was one of a pulmonary diffusion problem, not a ventilatory problem.


In spite of this, and despite the patient having a pO2 of 9 on high-flow oxygen the ITU registrar reduced the patient’s inspired oxygen concentration to an FiO2 of 0.28 (or 28%, as opposed to the 85% which is approximately what a non-rebreathe mask delivers). The patient’s oxygen saturations dropped to 80% again, and this poor gentleman was deprived of oxygen all because of this irrational and non-physiological appraisal of the nature of COPD.


The oxygen was of course then increased back up to high-flow, the patient taken for pressure support on ITU and thankfully survived to hospital discharge some 5 days later. Given my experience in similar cases I worry that without my perseverance in ITU referral the outcome may not have been so favourable.




Consider another recent case.  A respiratory arrest call – a patient with muscular dystrophy and type II respiratory failure with a severe respiratory acidosis and hypoxia. The attending ITU registrar tried to reduce this patient’s inspired oxygen concentration simply because they had type II respiratory failure. Here I had to speak up to ensure that this critically ill gentleman received the high-flow oxygen and ventilatory support that he needed. He had type II respiratory failure because his respiratory musculature was weakened by muscular dystrophy – this led to a severe hypoxia and acidosis.


This patient needed oxygen and ventilatory support, and so long as the arrest team is present there is literally zero risk of him losing his hypoxic drive.




In August I was called to a peri-arrest on the UK medical wards wherein a lady with a ‘history of COPD’ had been left with saturations of 80% on 5L of oxygen via nasal speculum. History and examination revealed a patient with a smoking history but no history of chronic cough or phlegm and an unlimited exercise tolerance (i.e. again poor evidence for COPD). On examination there were clinical features of collapse and consolidation of her whole right lung with excellent compensatory hyperventilation of her contralateral lung such that she had a respiratory alkalosis (i.e. her functioning left lung was so unaffected by any significant airflow limitation that it was managing to compensate for the total loss of her right lung).


I requested urgent ITU consultation and this lady received pressure support under their care and was later discharged home.





All critically unwell patients should have their hypoxia addressed with adequate oxygen therapy – to maintain saturations of equal to or above 95%, regardless of a COPD history. You should assess the patient clinically and if they are at risk of losing hypoxic drive you should establish a plan for ongoing monitoring (serial ABGs and clinical assessment) and appropriate titration of oxygen therapy.


Essentially the bottom line is that you shouldn’t apply a high-flow oxygen mask and then walk away. Sadly too many doctors are writing ‘COPD – Sats 88-92%’ in the notes and then walking away, which is arguably far more dangerous.


We should not withhold oxygen from any critically unwell patient without a good reason.  Of course if the patient is clinically well and stable it may be acceptable to maintain oxygen saturations of 88-92%.


Remember that patients are not labels and patients are not diseases. Patients can be complex; they have past medical histories and current medical problems. Sometimes these are related, sometimes not. Do not take disease labels at face value, and try not to believe with credulity everything that your colleagues told you.

Leave a Reply

Your email address will not be published. Required fields are marked *