Comparison of Three-Level Triage Versus Five-Level Emergency Severity Index for Prediction of Hospital Outcome of Emergency Patient.
- Author:
Sun Hwa SHIN
1
;
Sang Do SHIN
;
Hee Kang CHOI
;
Won Chul CHA
;
Young Sun RO
Author Information
1. Department of Emergency Medicine, Seoul National University Hospital, Seoul, Korea. shinsangdo@medimail.co.kr
- Publication Type:Original Article
- Keywords:
Triage;
Validity;
Outcome
- MeSH:
Ambulances;
Area Under Curve;
Demography;
Electronic Health Records;
Emergencies;
Emergency Medical Services;
Hospital Mortality;
Humans;
Insurance Coverage;
Intensive Care Units;
Korea;
North America;
ROC Curve;
Tertiary Care Centers;
Triage
- From:Journal of the Korean Society of Emergency Medicine
2011;22(5):478-488
- CountryRepublic of Korea
- Language:English
-
Abstract:
PURPOSE: This study compared the performance for the prediction of three-level versus five level triage tool (Emergency Severity Index version 4, ESI) on the hospital outcome of emergency patients. METHODS: This was an observational study that included all patients >15-year-of-age visiting an urban tertiary hospital emergency department (ED) in Korea from July 2007 to December 2008. We collected data from the electronic medical records, which included demographic factors, hospital outcome including admission to intensive care unit (ICU) and hospital mortality, and result of triage at arrival to ED. A three-level triage tool was used in the first year (July 2007 to June 2008), divided in four 3-month periods (threelevel phase). For 5 weeks, the three-level tool and ESI were used simultaneously (test phase). For the last 4 months, ESI was used for triage (five-level phase). We described the demographic findings of each study phase and compared the performance for the prediction of admission to ICU as well as hospital mortality, using the area under the receiver operating characteristic curve (AUC). RESULTS: A total of 34,342 patients during three-level phase, 3,371 during the test phase, and 11,048 during five-level phase were involved. Demographic findings about gender, age, week and time of visiting to ED, use of ambulance service, injury or illness, and results after ED management were similar without significant difference. Proportion of admission to ICU and hospital mortality at ED and on the ward was 6.1% in period 1, 6.1% in period 2, and 5.6% in period 3 of the three-level phase, 5.8% in the test phase, and 5.8% in the five-level phase. The calculated AUC of the three periods in the three-level phase was 0.747 (95% Confidence Interval, CI, 0.729~0.765), 0.786 (95% CI, 0.769~0.804), and 0.786 (95% CI, 0.769~0.804). During the test phase, the AUC of the three-level was 0.820 (95% CI, 0.786~0.854) and that of the five-level was 0.842 (95% CI, 0.809~0.874). During five-level phase using ESI, AUC of ESI was 0.826 (95% CI, 0.809~0.844). CONCLUSION: ESI showed greater ability to predict hospital mortality than the three-level triage tool of emergency patients in a Korean ED, where the emergency care system and insurance coverage differs from North America.
