Validated model predicts likelihood of SBI in pediatric ED patients

Access to the full content of this site is available only to registered healthcare professionals. Register to read more


  • Risk prediction model combining predefined clinical and biomarker variables (eg, irritability, dehydration, hypoxia, respiratory findings) may improve identification of serious bacterial infections (SBIs) in febrile children presenting to the emergency department (ED).

Why this matters

  • Acute SBIs are a primary reason for children presenting to the ED.
  • Prompt recognition of SBIs in children is essential for both ruling out the need for hospital admissions and preventing unnecessary antibiotic use, morbidity, and mortality.
  • A newly validated risk prediction model based on specific clinical/biomarker variables may help discriminate among pneumonia, other SBIs, and acute infection.

Key results

  • The biomarkers, procalcitonin and resistin, increased with pneumonia and SBIs vs no SBI (P<.001).
  • Similar to previous studies, SBIs increased with clinical variables such as work of breathing (OR, 10.4) and hypoxia (OR, 9.29).
  • A combination of clinical and biomarker variables (procalcitonin, resistin) improved discrimination of pneumonia and other SBIs (P=.03).
  • Improvement in concordance statistics after the addition of procalcitonin and resistin: pneumonia went from 0.88 to 0.90; other SBI models went from 0.82 to 0.84.

Study design

  • This prospective study analyzed variables in SBI diagnosis on 1101 children with fever (median age, 2.4 y) treated at the ED.
  • Funding: National Institute for Health Research.


  • Single-center study.

Coauthored with Chitra Ravi, MPharm