To the Editors:
The emergency management of hypoxaemic patients requires clinicians to avoid the hazard of dangerous hypoxaemia due to under-treatment with oxygen, whilst also avoiding the hazards of hypercapnic respiratory failure (iatrogenic hypercapnia) and oxygen toxicity, which may be caused by over-treatment with oxygen. In the past, many clinicians acted cautiously by giving high concentrations of oxygen to all potentially hypoxaemic patients and a culture evolved that “more is better”. However, since the 1960s, it has been known that some patient groups, particularly those with chronic obstructive pulmonary disease (COPD), are especially vulnerable to uncontrolled oxygen therapy and a recent randomised study showed that mortality in this patient group was doubled when high-concentration oxygen was used compared with controlled oxygen therapy [1–4]. It has also been demonstrated that hyperoxaemia is associated with increased mortality in patients with stroke, and in survivors of cardiac resuscitation and critically ill patients in the intensive care unit (ICU) [5–7]. The British Thoracic Society (BTS) guidelines for emergency oxygen use recommend a target oxygen saturation range of 94–98% for most emergency medical patients and a lower target range of 88–92% for those at risk of hypercapnic respiratory failure [8]. The proportion of emergency medical patients for whom each target range is appropriate is not known but a recent study of 1,022 emergency ambulance patients in the UK reported that 4% of ambulance patients had exacerbated COPD as the main diagnosis and 5.5% of non-COPD patients had saturation <90% at some time during the ambulance journey [9].
We studied an anonymised database consisting of blood gas analysis results from 3,524 specimens sent to the Biochemistry Laboratory of the Salford Royal University Hospital (Salford, UK) between June and November 2007. More than 95% of samples were from adult patients (neonatal and paediatric samples were analysed separately). About 5% of samples were arterialised earlobe capillary samples. Most emergency department and ICU samples were analysed on separate blood gas analysers and were not included in this analysis. The results are summarised in table 1. The saturation bands were chosen to correspond with the recommended target saturation ranges in the BTS guidelines, in addition to saturation bands above, below and between these ranges. The majority of patients were receiving oxygen therapy at the time of sampling. 2,693 samples had information about the inspired oxygen concentration, of which 19% were said to be breathing air at the time of sampling and 81% were on oxygen therapy ranging 24–100%.
More than a quarter of samples (26.9%) demonstrated hypercapnia with carbon dioxide tension (PCO2) >6.0 kPa or 45 mmHg consistent with type 2 respiratory failure, but only 198 (5.6%) samples had evidence of type 1 respiratory failure (oxygen tension (PO2) <8.0 kPa or 60 mmHg with normal PCO2). Detailed results are summarised in table 1.
1,458 (41.3%) samples had oxygen saturation >98% and 1,074 samples (30% of all samples) were hyperoxaemic with PO2 >15.0 kPa (112 mmHg), which is the upper limit of normal in our laboratory at the Salford Royal University Hospital. 362 (10%) samples were grossly hyperoxaemic with PO2 ≥20 kPa (≥150 mmHg). 362 (10.2%) samples had oxygen saturation <90% but only 2.7% were severely hypoxaemic with oxygen saturation <80%, and the majority of these samples exhibited hypercapnia with respiratory acidosis, indicating a need for either cautious oxygen therapy or ventilatory support.
The oxygen saturation was above the recommended target range for hypercapnic patients (saturation >92%) in 72% of our hypercapnic samples. 21.5% of all samples had findings consistent with respiratory acidosis (6.8% uncompensated and 14.7% compensated; table 1). Of the 239 patients with uncompensated respiratory acidosis, 59% had a normal base excess measurement consistent with acute respiratory acidosis and 41% had an elevated base excess (>2.0 mmol·L−1) consistent with chronic respiratory acidosis, although a proportion of these patients would have had acute decompensation of chronic respiratory acidosis.
Although it can be assumed that blood gas samples are usually taken from hospital patients at the highest risk of hypoxaemia, these results demonstrate that severe uncorrected hypoxaemia was relatively uncommon in this large database compared with the high prevalence of iatrogenic hyperoxaemia and hypercapnia, both of which may be associated with increased mortality [1–8]. Because the data were anonymised, the inspired oxygen concentration could not be validated and it was not possible to review the results by disease group or by outcome. For example, it is likely that some of the most hypoxaemic and acidotic specimens were collected during unsuccessful cardiac arrest procedures or from patients who were close to the end of their life. Additionally, it is likely that there were many repeat blood gas samples from the sickest patients with low blood oxygen levels.
This preliminary study suggests that severe uncorrected hypoxaemia with oxygen saturation <80% is relatively uncommon in modern hospital practice (2.7% of samples in this survey) and 54 of the 96 severely hypoxaemic samples had respiratory acidosis, whilst only seven of these samples had evidence of metabolic acidosis, which might be an expected consequence of severe hypoxaemia. Although severe hypoxaemia was uncommon, hyperoxaemia (30% of samples), hypercapnia (27% of samples) and respiratory acidosis (21% of samples) were relatively common findings and we found that hypercapnic (type 2) respiratory failure was five times more common than pure hypoxaemia (type 1 respiratory failure). These findings suggest that oxygen needs to be used with more caution in hospitals, particularly as it has recently been reported that hyperoxaemia in ICUs (arterial oxygen tension Pa,O2 >16 kPa or 120 mmHg) was associated with higher mortality than hypoxaemia (Pa,O2 <9 kPa or 67 mmHg) [7]. Our findings do not support the view of some clinicians that more patients are at risk of harm from lack of oxygen than from too much oxygen [10]. We now plan to seek permission to conduct a prospective study, which will involve review of diagnosis and outcome for each patient.
Footnotes
Statement of Interest
None declared.
- ©ERS 2012