Introduction Fever without source (FWS) in young children remains a difficult diagnostic problem, because clinical signs and symptoms are unreliable predictors of a serious bacterial infection (SBI). The aim of our study was to develop a risk index score based on objective laboratory values in children with FWS.

Methods Analysis of data collected from two consecutive prospectively enrolled cohorts of children with FWS was performed. Clinical assessment, white blood cell (WBC) count, urine analysis, C-reactive protein (CRP) and procalcitonin were analysed for all children. Univariate and multivariate logistic regression was used to identify predictive variables of SBI. A risk index score was developed with markers significantly associated with SBI. Two thirds of the population was used as a derivation set and one third as a validation set.

Results The data of 202 children were reviewed. In multivariate analysis, procalcitonin (odds ratio (OR) 37.6, 95% CI 5.8 to 243), CRP (OR 7.8, 95% CI 2 to 30.4) and urine dipstick (OR 23.2, 95% CI 5.1 to 104.8) remained significantly associated with SBI. White blood cell count and left shift were not independent predictors of SBI. Our risk index score, attributing points for values of procalcitonin and CRP superior to cut-off or for a positive urine dipstick, showed a sensitivity and a specificity of 94% and 81% for the identification of SBI, respectively. In the validation set comparable performances were obtained.

Conclusions Our laboratory risk index score could be a useful tool in the management of young children with FWS.