A model combining procalcitonin , c-reactive protein and urinalysis is superior to independent variables for predicting serious bacterial infections in febrile children

Introduction and aim. Fever is a common complaint for children addressing the emergency departments (ED). Distinguishing between febrile children with self limited viral infection and those with serious bacterial infection (SBI) may be challenging for practitioners, especially in younger population. Recently, a laboratory score, named the Lab-score, combining Procalcitonin, C-reactive protein and urine dipstick was developed for predicting SBI in febrile children. The Labscore was further assessed and validated in several studies. We aimed to search current literature and evaluate its value per se and by comparison with independent variables. Method. We search electronically the literature, in Medline, Embase and Google Scholar and identified the articles directly related to the Lab-score. We analysed the results of each study selected and corroborated the data. Results. The search returned 773 articles, six of them being relevant for the study. The highest sensitivity for the Lab-score for predicting SBI was 94% (95%CI: 82-99) and the highest specificity was 95% (95%CI: 93-96). The highest performance found for the Lab-score was reflected by an AUC of 0.91 (95%CI: 0.87-0.93) and the lowest by an AUC of 0.73 (95%CI: 0.69-0.77). We found that in four studies the Lab-score performed significantly better than independent predictors associated with SBI. Two studies found similar prediction power comparing the Lab-score with independent variables, one assessing a small group of infants and one assessing a much broader age group than all other studies. Conclusions. The Lab-score is a valuable tool for predicting SBI in febrile children addressing to ED and superior to independent variables, particularly in younger groups. Further validations are required for stronger conclusions.