Prediction Rule for Serious Infection Accurate in Low-Risk Febrile Infants

Mother taking temperature of infant
Study validates accurate prediction rule for identifying febrile infants age <60 days at low risk for serious bacterial infections using urinalysis, absolute neutrophil count, and procalcitonin levels.

An accurate prediction rule to identify febrile infants aged 60 days and younger at low risk for serious bacterial infections using urinalysis, absolute neutrophil count (ANC), and procalcitonin levels was validated according to the results of a prospective observational study published in JAMA Pediatrics.

The prediction rule was derived from data gathered from a random sample of 908 infants and validated in 913 infants. The prediction rule was based on clinical and laboratory data such as patient demographics, fever height and duration, clinical appearance, white blood cell count, ANC, serum procalcitonin, and urinalysis. It was derived and validated using binary recursive partitioning analysis.

Serious bacterial infections occurred in 9.3% of the 1821 infants, including 1.4% with bacteremia, 8.3% with urinary tract infections 0.5% with bacterial meningitis, and 0.9% with concurrent serious bacterial infections. Infants at low risk for serious bacterial infection were identified by the prediction rule using a negative urinalysis result, ANC ≤4090/µL, and serum procalcitonin ≤1.71 ng/mL. The sensitivity of the rule in the validation cohort was 97.7% (95% CI, 91.3%-99.6%), specificity was 60.0% (95% CI, 56.6%-63.3%), negative predictive value was 99.6% (95% CI, 98.4%-99.9%), and negative likelihood ratio was 0.04 (95% CI, 0.01%-0.15%). Of the infants included in the study, 3 were misclassified: 1 with bacteremia and 2 with a urinary tract infection. The rule did not miss any patients with bacterial meningitis and the performance was identical with the outcome restricted to bacteremia and/or bacterial meningitis; the same infant with bacteremia was missed in the latter.

Patient enrollment was based on research coordinator availability, but the study investigators believe the sample was representative because rates of specific serious bacterial infections were similar to those in other reported populations. This study also did not evaluate biomarkers other than procalcitonin, and did not investigate viral testing in the prediction rule.

According to the researchers, procalcitonin has superior test characteristics for bacteremia and bacterial meningitis compared with other biomarkers, and identification of viral pathogens diminishes but does not eliminate the risk for SBI in young febrile infants. Further, viral tests were not part of the protocol or routinely performed at study sites. The sample also included only 30 patients with bacteremia or bacterial meningitis, leading researchers to conclude that validation in cohorts with greater numbers of invasive infections is needed before implementation.

Related Articles

Until further validation of the prediction rule, the investigators warned that “clinicians must remain most cautious with infants younger than 28 days, in whom the risks of bacteremia and bacterial meningitis as well as herpes encephalitis are the greatest.” However, once further validated they believe that implementation of the rule, “has the potential to substantially decrease the use of lumbar punctures, broad-spectrum antibiotics, and hospitalization for many febrile infants [age] 60 days and younger.”


Kuppermann N, Dayan PS, Levine DA, et al. A clinical prediction rule to identify febrile infants 60 days and younger at low risk for serious bacterial infections [published online February 18 2019]. JAMA Pediatr. doi:10.1001/jamapediatrics.2018.5501