Background: Hypoxaemia is a predictor of pneumonia-related mortality. WHO published recommendations for oxygen therapy based on clinical signs which state that, when oxygen is plentiful, it should be given to children with central cyanosis, inability to drink, severe chest indrawing, RR >70 breaths/min, grunting with every breath (in young infants) or those who display head nodding. These guidelines, however, are based on a few studies only.
Aim: To assess the accuracy of combinations of clinical signs which predict hypoxaemia in pre-school children aged 2 months to 5 years with acute respiratory distress in hospitals in Switzerland and Senegal.
Methods: This observational study was conducted in four emergency units, two in Switzerland and two in Senegal. Patients aged 2 months to 5 years with acute respiratory distress were eligible for inclusion. Clinical signs were compared with transcutaneous blood saturation levels (SaO2).
Results: About 111 children were assessed, 67 in Switzerland and 44 in Senegal. The prevalence of hypoxaemia was 13%. Twelve models of combined symptoms were analysed. The WHO model, for when oxygen supply is ample, had the highest diagnostic performance with a sensitivity of 0.93 and a specificity of 0.60.
Conclusions: Clinical signs alone are unreliable for the detection of hypoxaemia. The current WHO model, for ample oxygen supply proved to be the best clinical predictor, although a great number of non-hypoxaemic children were unnecessarily treated because of the low specificity of this model.
Keywords: Hypoxaemia; acute respiratory distress; clinical signs; diagnostic accuracy; pulse oximetry; under-5 children.