Background and Significance
Most inpatient pediatric arrests are preventable by early recognition/treatment of deterioration. Early warning scoring (EWS) systems provide early identification of children at arrest risk. In 2009, the Children’s Hospital Early Warning Score (CHEWS) was developed and validated in a single center of pediatric cardiac patients.
Purpose
To further validate the Children’s Hospital Early Warning Score tool and algorithm in inpatient pediatric non- cardiac patients.
Methods
Nurses assess and document patients’ CHEWS scores during routine vital signs. An escalation of care algorithm directs either: routine care (score 0-2), increased assessment/intervention (3-4), or ICU consult/transfer (≥5). Sensitivity and specificity were estimated from a retrospective review of patients admitted to our inpatient units over 12 months who experienced arrest or unplanned ICU transfer (n=360) and a randomized comparison sample (n=776) of admissions. All patients in non-ICU or critical care areas that experienced an unplanned arrest or ICU transfer were included, patients at end of life with anticipated death were excluded from the case sample. All patients admitted to non-ICU or critical care were considered for inclusion for comparison control sample.
Analysis
The previously validated Pediatric Early Warning Score (PEWS) tool was used for comparison. Patients’ highest CHEWS scores were compared to calculated PEWS scores. Area under the receiver operating characteristic (AUROC) curve was calculated for PEWS and CHEWS to measure discrimination.
Findings
CHEWS algorithm sensitivity was 97.8 (≥2), 84.2 (≥4) and 75.6 (≥5) versus PEWS of 82.8 (≥2), 54.4 (≥4), and 38.9 (≥5). CHEWS specificity was 52.5 (≥2), 80.9 (≥4), and 88.5 (≥5) versus PEWS of 63.7 (≥ 2), 85.3 (≥4) and 93.9 (≥5). The AUROC curve for CHEWS was 0.902 compared to PEWS 0.798.
Implications
In this single center examination, the CHEWS demonstrated a higher discrimination and sensitivity than the PEWS in identifying deterioration in hospitalized children.
Article info
Identification
Copyright
© 2014 Published by Elsevier Inc.
