Clinical and laboratory features associated with mortality in children with severe malnutrition in Papua New Guinea.

Background: The World Health Organization has a clinical and syndromic approach to the management of severe acute malnutrition which recognises that laboratory investigations are often not possible where children with severe malnutrition present. In low- and middle-income countries including Papua New Guinea, rates of death from severe malnutrition in many hospitals remain 10% or more.Aim: To evaluate the clinical predictors of death and the association between disturbances of electrolytes and haematological investigations in children with severe malnutrition and the risk of mortality.Methods: The clinical and laboratory predictors of death in a prospective cohort of 150 children with severe malnutrition admitted to a provincial hospital in Papua New Guinea were analysed. The clinical signs and electrolytes, complete blood count and liver function tests at presentation and on Days 3 and 5 were recorded.Results: The strongest independent predictors of mortality at assessment on admission were a low child Glasgow coma scale (≤12), hypoxaemia (SpO₂ <90%), prolonged capillary refill (>3 seconds) and dysnatraemia (<130 or >150 mmol/L). The area under the receiver operating characteristics curve for these four variables was 0.93.Conclusions: That three of these four criteria correspond closely to the WHO Emergency Clinical Signs reinforces the value of a system of triage and risk assessment in children with severe malnutrition. If a child has emergency signs they should be managed in an area on the ward where close monitoring and supportive care can be provided, the WHO guidelines for severe malnutrition followed, and other specific care provided. Measurements of serum sodium, particularly in children with diarrhoea and dehydration, is also important in risk assessment and management.