Evaluation of a Pediatric Early Warning Score as a Predictor of Occult Invasive Bacterial Infection in the Pediatric Emergency Department

Pediatr Emerg Care. 2022 May 1;38(5):195-200. doi: 10.1097/PEC.0000000000002554. Epub 2021 Oct 28.

Abstract

Objectives: The aims of the study were to evaluate the diagnostic performance of Pediatric Early Warning Score (PEWS) to predict occult invasive bacterial infection (IBI) in well-appearing pediatric emergency department (PED) patients without known risk factors for bacterial infection and to compare PEWS to heart rate (HR) and Emergency Severity Index (ESI).

Methods: We performed a retrospective case-control analysis of febrile PED patients aged 60 days to 18 years over a 2-year period. Subjects were excluded if they were ill appearing, admitted to an intensive care unit, or had a known high-risk condition. Cases of occult IBI were included if they had a noncontaminant positive culture other than an isolated positive urine culture. Two febrile control subjects were identified for each case. Odds ratios and receiver operating characteristic curves were evaluated to determine performance characteristics of PEWS at triage and disposition, age-adjusted HR at triage and disposition, and ESI at triage.

Results: Compared with 178 controls, 89 cases had higher disposition PEWS, higher disposition HR, lower ESI, and higher rate of hospital admission. Disposition PEWS ≥3 (odds ratio, 2.57; 95% confidence interval, 1.08-6.18), disposition HR > 99th percentile, and ESI demonstrated increased odds of occult IBI. Area under the receiver operating characteristic curve for disposition PEWS (0.56) was similar to triage PEWS (0.54), triage HR (0.54), disposition HR (0.58), and ESI (0.65).

Conclusions: Subjects with PEWS ≥3 at PED disposition have increased odds of occult IBI; however, PEWS has poor discriminative ability at all cutoffs. We cannot recommend PEWS used in isolation to predict occult IBI.

MeSH terms

  • Bacterial Infections* / diagnosis
  • Child
  • Early Warning Score*
  • Emergency Service, Hospital
  • Humans
  • ROC Curve
  • Retrospective Studies
  • Triage