To investigate the severity and duration of hypoxaemia in 703 children with severe or very severe pneumonia presenting to Goroka Hospital in the Papua New Guinea highlands; to study the predictive value of clinical signs for the severity of hypoxaemia, the predictive value of transcutaneous oxygen saturation (SpO2) and other variables for mortality.

Prospective evaluation of children with severe or very severe pneumonia. SpO2 was measured at the time of presentation and every day until hypoxaemia resolved. Children with a SpO2 less than 85% received supplemental oxygen. By comparing with a retrospective control group for whom oxygen administration was guided by clinical signs, we evaluated whether there was a survival advantage from using a protocol for the administration of oxygen based on pulse oximetry. We determined normal values for oxygen saturation in children living in the highlands.

In 151 well, normal highland children, the mean SpO2 was 95.7% (SD 2.7%). The median SpO2 among children with severe or very severe pneumonia was 70% (56–77); 376 (53.5%) had moderate hypoxaemia (SpO2 70–84%); 202 (28.7%) had severe hypoxaemia (SpO2 50–69%); and 125 (17.8%) had very severe hypoxaemia (SpO2 2 on presentation, severe malnutrition, measles and history of cough for more than 7 days. The mortality risk ratio between the 703 children managed whose oxygen administration was guided by the use of pulse oximetry and the retrospective control group who received supplemental oxygen based on clinical signs was 0.65 (95%CI 0.41–1.02, two-sided FisherÕs exact test, P = 0.07).

There is a need to increase the availability of supplemental oxygen in smaller health facilities in developing countries, and to train health workers to recognise the clinical signs and risk factors for hypoxaemia. In moderate sized hospitals a protocol for the administration of oxygen based on pulse oximetry may improve survival.