Development and validation of a Paediatric Early Warning Score for use in the emergency department: a multicentre study

Background Paediatric Early Warning Scores (PEWSs) are being used increasingly in hospital wards to identify children at risk of clinical deterioration, but few scores exist that were designed for use in emergency care settings. To improve the prioritisation of children in the emergency department (ED), we developed and validated an ED-PEWS. Methods The TrIAGE project is a prospective European observational study based on electronic health record data collected between Jan 1, 2012, and Nov 1, 2015, from five diverse EDs in four European countries (Netherlands, the UK, Austria, and Portugal). This study included data from all consecutive ED visits of children under age 16 years. The main outcome measure was a three-category reference standard (high, intermediate, low urgency) that was developed as part of the TrIAGE project as a proxy for true patient urgency. The ED-PEWS was developed based on an ordinal logistic regression model, with cross-validation by setting. After completing the study, we fully externally validated the ED-PEWS in an independent cohort of febrile children from a different ED (Greece). Findings Of 119 209 children, 2007 (1·7%) were of high urgency and 29 127 (24·4%) of intermediate urgency, according to our reference standard. We developed an ED-PEWS consisting of age and the predictors heart rate, respiratory rate, oxygen saturation, consciousness, capillary refill time, and work of breathing. The ED-PEWS showed a cross-validated c-statistic of 0·86 (95% prediction interval 0·82–0·90) for high-urgency patients and 0·67 (0·61–0·73) for highurgency or intermediate-urgency patients. A cutoff of score of at least 15 was useful for identifying high-urgency patients with a specificity of 0·90 (95% CI 0·87–0·92) while a cutoff score of less than 6 was useful for identifying low-urgency patients with a sensitivity of 0·83 (0·81–0·85). Interpretation The proposed ED-PEWS can assist in identifying high-urgency and low-urgency patients in the ED, and improves prioritisation compared with existing PEWSs.