The ability of early warning scores (EWS) to detect critical illness in the prehospital setting: A systematic review

Abstract Aim To examine whether early warning scores (EWS) can accurately predict critical illness in the prehospital setting and affect patient outcomes. Methods We searched bibliographic databases for comparative studies that examined prehospital EWS for patients transported by ambulance in the prehospital setting. The ability of the different EWS, including pre-alert protocols and physiological-based EWS, to predict critical illness (sensitivity, odds ratio [OR], area under receiver operating characteristic [AUROC] curves) and hospital mortality was summarised. Study quality was assessed using the Newcastle–Ottawa Scale. Results Eight studies were identified. Two studies compared the use of EWS to standard practice using clinical judgement alone to identify critical illness: the pooled diagnostic OR and summary AUROC for EWS were 10.9 (95%CI 4.2–27.9) and 0.78 (95%CI 0.74–0.82), respectively. A study of 144,913 patients reported age and physiological variables predictive of critical illness: AUROC in the independent validation sample was 0.77, 95% CI 0.76–0.78. The high-risk patients stratified by the national early warning score (NEWS) were significantly associated with a higher risk of both mortality and intensive care admission. Data on comparing between different EWS were limited; the Prehospital Early Sepsis Detection (PRESEP) score predicted occurrence of sepsis better than the Modified EWS (AUROC 0.93 versus 0.77, respectively). Conclusion EWS in the prehospital setting appeared useful in predicting clinically important outcomes, but the significant heterogeneity between different EWS suggests that these positive promising findings may not be generalisable. Adequately powered prospective studies are needed to identify the EWS best suited to the prehospital setting.