Introduction: In-hospital Cardiac Arrest (IHCA) has increasingly been recognized as a separate entity from out-of-hospital Cardiac Arrest with regard to epidemiology, clinical prediction, and outcomes. The incidence of adult IHCA was about 1 per 1,000 bed-days in the US and 15 to 20% of these patients survived to hospital discharge. Despite the morbidity and mortality, clinical tools for predicting IHCA are scarce, particularly in the Emergency Department (ED).
Hypothesis: Using Electronic Medical Record (EMR) data, we sought to include patients presenting to our ED to 1) describe the incidence of ED IHCA and 2) develop and validate a triage tool for predicting ED IHCA.
Methods: This retrospective cohort study used EMR data from a tertiary teaching hospital with approximately 100,000 ED visits per year. We extracted data from 741,795 ED visits over a 7-year period (Jan 1, 2009 to Dec 31, 2015). For repeat visits, we randomly selected one visit per patient. Only adult patients were included in this analysis. Patient demographics and triage information including triage levels, vital signs (temperature, pulse rate, systolic and diastolic blood pressure, respiratory rate, and oxygen saturation) and mental status (coded as Glasgow Coma Scale) were extracted as potential predictors. The primary outcome, ED IHCA, was identified via a resuscitation code. The predictive tool was developed in 60% of the data and validated in the remaining 40%.
Results: A total of 330,355 adult ED patients were included during the 7-year study period. Of them, 916 (0.3%) developed ED IHCA. The triage predictive tool, including age, sex, triage levels, and triage vital signs with cutoffs similar to those in published early warning scores, showed excellent discrimination (area under the receiver operating characteristic [AUROC] curve, 0.90) and calibration (P=0.30 for Hosmer-Lemeshow [HL] test). When applied to the validation cohort, it maintained good discriminatory ability (AUROC, 0.87) and calibration (P=0.17 for HL test).
Conclusions: IHCA within the ED is not uncommon. We developed and validated a novel tool in predicting imminent IHCA events in the ED. Implementation of this tool may help identify high-risk patients and reduce potentially preventable deaths.