Diagnostic Accuracy of Simplified Prehospital Sepsis Screening Tools for Identifying Sepsis and Predicting 30-Day Mortality: A Retrospective Diagnostic Accuracy Study

Thongpitak Huabbangyang,1 Gawin Tiyawat,1 Nitiporn Puysopa,1 Thapanee Wattanasaringkarn,1 Jiratchaya Khunjan,1 Warinthorn Inthanon,1 Nattanicha Duangkaew,1 Chanathip Wanphen,2 Adisak Nithimathachoke31Department of Disaster and Emergency Medical Operation, Faculty of Science and Health Technology, Navamindradhiraj University, Bangkok, Thailand; 2Division of Emergency Medical Service and Disaster, Faculty of Medicine Vajira Hospital, Navamindradhiraj University, Bangkok, 10300, Thailand; 3Department of Emergency Medicine, Faculty of Medicine Vajira Hospital, Navamindradhiraj University, Bangkok, ThailandCorrespondence: Adisak Nithimathachoke, Email adisak@nmu.ac.thBackground: Early recognition of sepsis in prehospital settings remains challenging because emergency medical services (EMS) clinicians must make rapid triage decisions using limited physiological information before hospital arrival. Several sepsis screening tools have been validated in emergency department settings; however, comparative evidence in prehospital populations remains limited.Purpose: This study evaluated the diagnostic and prognostic performance of five simplified screening tools—mNEWS, qSOFA, mSIRS, mSOS, and MEWS—for identifying sepsis and predicting 30-day mortality in prehospital patients.Methods: This retrospective diagnostic accuracy study included 268 adult patients (≥ 18 years) transported by Vajira Emergency Medical Service, Bangkok, Thailand, to Vajira Hospital between January 2018 and October 2024. Prehospital clinical parameters recorded by paramedics and emergency nurse practitioners were analyzed against the final hospital diagnosis of sepsis and 30-day in-hospital mortality. Predefined thresholds (mSOS ≥ 4, MEWS ≥ 5, mNEWS ≥ 5, qSOFA ≥ 2, and mSIRS ≥ 2) were used for the primary analyses. Diagnostic performance was evaluated using the area under the receiver operating characteristic curve (AuROC), sensitivity, and specificity.Results: Among 268 patients, MEWS demonstrated the highest discrimination for identifying sepsis (AuROC 0.84; sensitivity 64.9%; specificity 79.1%). mSOS (AuROC 0.83) and mSIRS (AuROC 0.82) demonstrated high sensitivity (> 96%) but lower specificity (~60%), while mNEWS showed the highest sensitivity (99.3%) but poor specificity (47.0%). qSOFA demonstrated the lowest discrimination (AuROC 0.72). For predicting 30-day mortality, MEWS demonstrated the highest performance (AuROC 0.70; sensitivity 72.2%; specificity 64.5%), followed by mNEWS and mSOS (AuROC 0.69 each), whereas qSOFA (AuROC 0.64) and mSIRS (AuROC 0.60) demonstrated lower discrimination.Conclusion: MEWS demonstrated the most balanced diagnostic performance for identifying sepsis and predicting 30-day mortality in prehospital patients. Highly sensitive tools may be useful for early screening, whereas MEWS may better support risk stratification. Further external validation is required before broader implementation.Keywords: sepsis, emergency medical services, prehospital care, diagnostic accuracy, prognosis, mortality