Introduction Some clinicians take an aggressive approach to supplemental oxygen therapy to avoid the dangers of hypoxaemia but many patients, especially those with COPD, are at risk from uncontrolled oxygen therapy. Furthermore, recent publications have suggested an independent association between hyperoxaemia and in-hospital mortality of ICU patients (de Jonge 2008, Kilgannon 2010).

Methods We studied a 1-year database of 29 587 hospital blood gas specimens including samples from A&E and ICU. Only the first adequate sample from each patient was analysed; (n=7956). Samples were grouped in saturation bands corresponding to the recommended target saturation ranges in the BTS Emergency Oxygen Guideline and saturation bands above, below and between these ranges (Abstract P69 table 1). A random sub-sample of 360 specimens was analysed in more detail for clinical outcomes and the prevalence of risk factors for type 2 respiratory failure.

Results Excluding A&E cases, only 10% of blood gas samples demonstrated Type 1 respiratory failure with PO₂ <8 kPa (60 mm Hg) and normal CO₂ levels. Thirty per cent of first A&E samples had Type 1 respiratory failure. Of 7965 samples analysed, 22% were hypercapnic with PCO₂ >6 kPa (45 mm Hg) consistent with Type 2 respiratory failure and the oxygen saturation was above 92% in 54% of these hypercapnic patients. 22% of all samples were hyperoxaemic with PO₂>15 kPa (112 mm Hg) and 26% had saturation >98%. Within the sub-sample of 360 specimens there were 72 in-hospital deaths (20%). The lowest mortality was 8% in the 88–92% saturation range and the highest mortality was 39% in those with saturation <88%. Of 85 patients documented to be at risk of Type 2 respiratory failure, 62 (73%) had oxygen saturations greater than the 88–92% target range recommended in BTS Guidelines.

Conclusions Despite the introduction of the BTS Guidelines, hyperoxaemia is still a common finding in blood gas specimens in 2011. Type 2 respiratory failure is twice as common as Type 1 failure except for A&E samples but the majority of patients at risk of Type 2 respiratory failure had oxygen saturations above the recommended target range. These results demonstrate the need for a prospective study of the link between oxygen therapy and overall hospital mortality.