RESUMEN
To determine utility of procalcitonin (PCT) for the prediction of bacterial infection in critically ill children, we analysed the relationship between serum PCT, cultures and other laboratory markers of bacterial sepsis or viral infection in a tertiary paediatric intensive care unit (PICU). The outcome measures were levels of PCT in proven bacteraemia, pneumonia and viral respiratory infection; and comparison of PCT to immature to total neutrophil ratio (ITR) in prediction of bacteraemia. In 420 children with suspected sepsis, 1,226 serum PCT levels were analysed. Children with bacteraemia had a higher median PCT (2.03 ng/ml, interquartile range [IQR] 0.67-42.4) than those who did not have bacteraemia (0.82 ng/ml, IQR 0.295-2.87) (P=0.033). PCT was a significant but only moderate predictor of bacteraemia (AUC 0.65). In 866 episodes of suspected sepsis where paired PCT and ITR were performed, the median ITR in children with bacteraemia was 0.19 ng/ml (IQR 0.04-0.35), and the median PCT was 6.5 ng/ml (IQR 0.71-61.8). PCT was a marginally better predictor of bacteraemia (AUC 0.69) than the ITR (AUC 0.66). In children with viral respiratory tract infection only, the median PCT was 1.26 ng/ml (0.35-5.5), and in those with likely bacterial pneumonia the median PCT was 0.80 ng/ml (IQR 0.28-1.70). In a heterogeneous population of children in a tertiary PICU, PCT measured at a single timepoint was a moderate predictor of proven bacteraemia. In our population PCT did not reliably identify localised bacterial infection or distinguish bacterial from viral respiratory infection.