Background There is controversy about the optimal level of oxygenation for patients in intensive care units (ICU) and several recent publications have raised the possibility of harm from hyperoxaemia amongst ICU patients.¹ Most recent ICU publications have reported that hyperoxaemia is common. The aim of this project was to audit blood gas data at two hospitals to determine if there is evidence of changing practice in recent years.

Methods We audited all available ICU blood gas datasets for Hospital A for 2005 (n = 16,734), 2010 (n = 12,714) and 2015 (n = 17,292) and data from Hospital B from 2012–2013 (n = 11,006) and for 2015 (n = 22,223).

Results At Hospital A, the percent of ICU blood gas samples with hyperoxaemia (SaO₂ > 98%) fell from 57.4% in 2005 to 45.1% in 2010 and 29.0% in 2015. Mean SaO₂ fell from 97% in 2005 to 96% in 2015. The mean PaO₂ also fell from 15.1 KpA in 2005 to 13.5 in 2015 (The reference range for PaO2 is 12.0 to 15.0 kPa). Samples from hospital B did not span such a wide time range but they demonstrated a similar fall in the proportion of hyperoxaemic samples from 42.7% in 2012–2013 to 29.7% in 2015 which was very similar to the data for Hospital A in 2010 and 2015 (45.1% and 29.0%).

Conclusions There is evidence from two hospitals that the use of oxygen in Intensive Care Units has become more conservative in recent years, possibly as a result of recent publications which have identified poor outcomes in association with hyperoxaemia. A wider survey will be undertaken to determine if this trend is taking place in Intensive Care Units throughout the UK.

Reference

1. Helmerhorstet al. Crit Care Med 2015;43:1508–19.