Objective: This study describes the predictors of in-hospital cardiac arrest (IHCA) within 24 h of ED triage and evaluates their ability to predict patients at risk of IHCA.
Methods: A case-control study was conducted in the ED. 'Cases' are herein defined as hospitalised patients who experienced IHCA within 24 h after ED triage. The exclusion criteria were those younger than 16 years old, cases of traumatic arrest, or had do-not-resuscitate orders. The controls were adults, non-traumatic cases, who did not experience IHCA within 24 h of ED triage. A multivariable regression model was used to identify significant predictors of IHCA. The ability to discriminate was quantified by utilising an area under receiver operating characteristic (AuROC) curve.
Results: Two hundred and fifty IHCAs were compared with 1000 controls. Five predictors emerged that were: higher National Early Warning Score (NEWS) at triage, equal or increase of NEWS after ED management, coronary artery disease as a comorbid disease, the use of a vasoactive agent, and initial serum bicarbonate level lower than 23.5 mmoL/L, independently associated with IHCA. The AuROC of the final model from all predictors was 0.91 (95% CI 0.89-0.93) higher than NEWS alone model (AuROC at 0.78, 95% CI 0.74-0.81).
Conclusions: We conclude that a combination of NEWS and four independent predictors identify patients at risk of IHCA more effectively than NEWS alone.
Keywords: ED; Early Warning Score; National Emergency Warning Score; in-hospital cardiac arrest; predictor.
© 2019 Australasian College for Emergency Medicine.