RESUMO
AIM: To measure the prognostic value of C-reactive protein (CRP) and its ability to predict pneumonia-associated complications. METHODS: A 3.75-years retrospective cohort analysis of all paediatric emergency department visits with a discharge diagnosis of pneumonia. Visits where CRP was not measured or with a discharge diagnosis of viral pneumonia were excluded. The following five outcomes were studied: hospitalisation, presence of parapneumonic effusion (PPE), placement of a chest drain, admission to paediatric intensive care unit (PICU) and bacteremia. A multivariate model was constructed and validated using k-fold cross-validation. RESULTS: During the study time period, there were 2561 visits for pneumonia, of which 810 were included in our analysis. The median age of included children was 3.2 years (range 0.2-17.7). Overall, 38.8% visits ended in hospitalisation, 2.2% required admission to PICU, 15.2% were complicated by a PPE of which 28% required the placement of a chest drain. Statistically significant association was found between CRP levels and each of these outcomes (P < .001). Incorporating CRP within a multivariate prediction model provided an area under the curve of up to 0.96. CONCLUSION: CRP can be a useful prognostic marker when evaluating a patient with suspected bacterial pneumonia and could help the paediatrician in identifying patients needing closer follow-up.
