Predicting Critical Illness in the Prehospital Setting: A Comparison of the New Zealand Early Warning Score (NZEWS) and the Current Clinical Triage Status Codes

Abstract

Emergency Ambulance Service (EAS) personnel often represent the first healthcare contact for acutely unwell patients. They work in complex and unpredictable environments and make clinical decisions with limited information and support. Early identification of patients at risk of critical illness can reduce the time to definitive treatment and improve patient outcomes. However, the difficulty in identifying deteriorating patients in the prehospital setting has been well documented.

Early Warning Scores (EWS) have been developed to help identify patients at risk of deterioration and are routinely used in the hospital setting. In Aotearoa New Zealand (AoNZ), the severity of a prehospital patient’s condition is currently categorised using the clinical triage Status Codes based on EAS personnel’s clinical judgement. The New Zealand Early Warning Score (NZEWS) is used in AoNZ hospitals and its adoption into the prehospital setting could potentially increase the early identification of at-risk patients.

This thesis aims to establish whether the NZEWS has a higher diagnostic accuracy than the current clinical triage Status Codes for predicting critical illness in the AoNZ prehospital setting.

Firstly, a systematic literature review was conducted to establish whether the use of EWS can predict critical illness in the prehospital setting. The 16 included studies evaluated the prehospital diagnostic accuracy of the Modified Early Warning Score (MEWS), the National Early Warning Score (NEWS), or the National Early Warning Score 2 (NEWS2) for predicting mortality. The prehospital use of EWS generally demonstrated a moderate to good diagnostic performance with an Area Under the Receiver Operating Characteristic curve ranging from 0.68 (95% CI: 0.64-0.73) to 0.90 (95% CI: 0.82-0.97), with a higher performance for short-term mortality. However, the high optimal cut-off point required for predicting short-term mortality suggested that EWS would be most useful for the critically ill, who make up a small proportion of the prehospital patient population.

Secondly, a retrospective observational cohort compared the prehospital use of the New Zealand Early Warning Score (NZEWS) with the current practice of EAS personnel allocating a clinical triage Status Code for predicting critical illness. Recent studies have used short-term mortality to measure critical illness and therefore the primary outcome for this research was 2-day mortality. The study included 125,278 patients with 1443 patients (1.2%) dying within 2 days. There was a strong correlation with 2-day mortality between both the prehospital NZEWS and the clinical triage Status Codes. The diagnostic accuracy of the initial NZEWS (AUROC 0.80, 95% CI: 0.79-0.81) was higher than the diagnostic accuracy of the initial clinical triage Status Code (AUROC 0.74, 95% CI:0.73-0.76) in predicting 2-day mortality. The optimal cut-off point for prehospital NZEWS was 6 (moderate risk) which suggested a greater utility for the prehospital patient population than previous research. Older patients (AOR 1.04, 95% CI: 1.03-1.04), Māori (AOR 1.34, 95% CI: 1.13-1.59), Pacific people (AOR 1.36, 95% CI: 106-1.76), and those living in rural locations (AOR 1.17, 95% CI: 1.02-1.33) were also associated with increased odds of 2-day mortality.

These findings demonstrate that Prehospital NZEWS may be a useful additional tool to help EAS personnel identify patients at risk of clinical deterioration. The NZEWS should be seen as a support to clinical decision-making and not as a replacement for structured assessment and clinical judgement. Its standardised approach may help bring consistency in the response to clinical deterioration and help communication across the healthcare setting. However, this standardisation may not meet the needs of different patient populations, and modifying the NZEWS could increase its performance further and help address health disparities.