Development and internal validation of a caller-reported triage decision-support tool for children aged 0‐4 years presenting with breathing problems

Introduction: The aim of this study was to develop and internally validate the Paediatric Respiratory Emergency Prioritisation (PREP) score, a caller-reported decision-support tool for children aged 0‐4 years presenting with breathing problems, and to explore its potential role in supporting clinician-led risk stratification during emergency calls. Methods: This service evaluation analysed routinely collected operational data from children aged 0‐4 years with Medical Priority Dispatch System (MPDS) code 06E01 (‘Breathing problems’) within the Welsh Ambulance Service University NHS Trust between September 2022 and May 2025. Caller-reported features (consciousness level, airway status, respiratory rate, signs of respiratory distress, relevant medical history and MPDS diagnostic suffix) were evaluated using logistic regression and structured into the PREP score ‐ a candidate, additive, decision-support score developed using pre-alert activation as an operational proxy for clinician-recognised urgency. Model development used a training dataset, with internal validation performed on an independent test dataset. Performance was assessed using discrimination, calibration and classification metrics. Results: A total of 5935 incidents were included. The PREP model demonstrated moderate discrimination (area under the curve 0.73), with good calibration (Brier score 0.093). At the optimised probability threshold (0.14) sensitivity was 53%, specificity 81%, precision 26% and negative predictive value 93%. The F1 score was 0.35. Retrospective application showed that 38% of children scored zero, indicating no high-risk indicators on the basis of caller-reported information. Conclusion: Within this service evaluation, the PREP score demonstrated promising internal performance as a clinician-facing decision-support tool for early risk stratification of young children with breathing problems during emergency calls. The high negative predictive value suggests potential utility for identifying children less likely to trigger pre-alert activation within existing clinician-led triage pathways. Prospective evaluation of downstream outcomes, safety and external validity is required before implementation.