Performance of Febrile Infant Algorithms by Duration of Fever

OBJECTIVES To analyze the performance of commonly used blood tests in febrile infants ≤90 days of age to identify patients at low risk for invasive bacterial infection (bacterial pathogen in blood or cerebrospinal fluid) by duration of fever. METHODS We conducted a secondary analysis of a prospective single-center registry that includes all consecutive infants ≤90 days of age with fever without a source evaluated at 1 pediatric emergency department between 2008 and 2021. We defined 3 groups based on caregiver-reported hours of fever (<2, 2–12, and ≥12) and analyzed the performance of the biomarkers and Pediatric Emergency Care Applied Research Network, American Academy of Pediatrics, and Step-by-Step clinical decision rules. RESULTS We included 2411 infants; 76 (3.0%) were diagnosed with an invasive bacterial infection. The median duration of fever was 4 (interquartile range, 2–12) hours, with 633 (26.3%) patients with fever of <2 hours. The area under the curve was significantly lower in patients with <2 hours for absolute neutrophil count (0.562 vs 0.609 and 0.728) and C-reactive protein (0.568 vs 0.760 and 0.812), but not for procalcitonin (0.749 vs 0.780 and 0.773). Among well-appearing infants older than 21 days and negative urine dipstick with <2 hours of fever, procalcitonin ≥0.14 ng/mL showed a better sensitivity (100% with specificity 53.8%) than that of the combination of biomarkers of Step-by-Step (50.0% and 82.2%), and of the American Academy of Pediatrics and Pediatric Emergency Care Applied Research Network rules (83.3% and 58.3%), respectively. CONCLUSIONS The performance of blood biomarkers, except for procalcitonin, in febrile young infants is lower in fever of very short duration, decreasing the accuracy of the clinical decision rules.