Study design and setting
This retrospective cohort study used data extracted from the National Emergency Department Information System (NEDIS), which has been operated by the National Emergency Medical Center of Korea since 2003. NEDIS collects data from all patients visiting the emergency room. Moreover, in South Korea, there is no registry that manages only trauma patients—NEDIS performs the trauma registry function.
The data from NEDIS contain patient demographics and clinical information, which were prospectively collected from 399 emergency medical facilities in Korea . All patient-related information is automatically transferred to a server managed by the central government within at least 14 days after the patient’s discharge. NEDIS is considered a dependable nationwide reference of emergency department (ED) information.
We retrospectively reviewed the data collected from NEDIS between January 2015 and December 2017. All trauma patients who visited EDs located in Gangwon province were included. We identified trauma patients on the basis of the International Classification of Disease (ICD) codes, which is included in the NEDIS data. Patients who 1) were not diagnosed with damage accident, 2) were aged < 15 years, 3) were not involved in an accident such as drowning or intoxication, 4) were transferred out after emergency medical care, 5) had a cardiac arrest before emergency medical care, and 6) had inadequate data for analysis were excluded from this study.
This study was approved by the institutional review board (IRB) of Wonju Severance Christian Hospital (CR319164). The requirement for the patients’ informed consent was waived by the IRB owing to the retrospective nature of the study.
Gangwon Province And Designation Of Trauma Centers In Korea
According to the resource and capacity of medical institutions, the emergency medical facilities in Korea are categorized into three types (regional emergency medical centers, local emergency medical centers, and local emergency medical service institutions) by the Ministry of Health and Welfare. In 2019, there were 35 regional emergency medical centers, 126 local emergency medical centers, and approximately 240 local emergency medical service institutions. Since 2012, 17 regional emergency medical centers were designated as RTCs, of which 14 were operational at the time of this.
Gangwon province has the second largest area among the eight provinces of Korea and is approximately 16,875 km2. However, its population and density are the lowest in Korea, at approximately 1.5 million and 90/km2, respectively. Geographically, the mountain region accounts for most of the area; therefore, patient transportation usually takes longer than that in other provinces. Previously, there was 1 regional emergency medical center designated as an RTC, 2 regional emergency medical centers, 4 local emergency medical centers, and 15 local emergency medical service institutions in the province. Wonju Severance Christian Hospital was designated as a regional emergency medical center in 2002 and was officially opened as an RTC in 2015 (Fig. 1).
Definition Of Variables
Using data extracted from NEDIS, we examined basic demographic variables, types of hospitals, mechanism of injury, vising routes, types of transportation, and other medical information. The ED visiting route was categorized into two types: direct visit and visit via transfer. The type of transportation included ambulance, helicopter, other vehicles, walking, etc. Medical information included vital signs and level of consciousness at the time of ED arrival, patient’s disposition after initial care at the ED, diagnosis based on the ICD codes, mortality, in-hospital death, and time factors such as injury time, ED visiting time, discharge time from ED, or after admission. Level of consciousness was divided into the following categories: alert, verbal response, painful response, and unresponsiveness. Using the time factors, we calculated the duration from injury to ED visit, length of stay at ED, and hospital length of stay. Moreover, the severity of the injury was assessed by calculating an ICD-based injury severity score (the International Classification of Disease-Based Injury Severity Score, ICISS) based on the diagnosis entered in NEDIS. The survival risk ratio was assigned on the basis of the individuals’ ICD codes, and ICISS was finally calculated on the basis of the survival risk ratio. We defined major trauma as ICISS < 0.9 based on previous results . Patients were analyzed according to the type of receiving hospital and then categorized into the RTC and non-RTC groups. The non-RTC group comprised regional emergency medical centers, local emergency medical centers, and local emergency medical service institutions.
To compare the characteristics of the patients who visited an RTC and those who had not, independent two-sample t test was used for continuous variables and chi-square test or Fisher’s exact test were used to compare categorical variables. Univariate logistic regression analysis was used to test the association between death and age, sex, mechanism of injury, visiting route, transportation, level of consciousness, systolic blood pressure, diastolic blood pressure, pulse rate, respiration rate, and oxygen saturation. Multivariate analysis was adjusted using the significant factors affecting death determined in univariate logistic regression analysis.
We also created a propensity score (PS)-matched cohort by attempting to match each patient who visited an RTC and non-RTC (a 1:2 match). To reduce the effects of potential confounding factors when comparing the prognosis of the patient between RTC and non-RTC, the PS was estimated using a multivariate logistic regression model with the group as the dependent variable and ICISS, age, and sex as covariates. A nearest neighbor matching algorithm with a “greedy” heuristic (one that always implements the best immediate or local solution) was used to match the patients. After matching, we also evaluated the degree of balance in measured covariates between the RTC and non-RTC. An additional file shows this in more detail [see Additional file 1].
All reported P values are two-sided, and P values < .05 were considered to indicate statistical significance. SAS software v 9.4 (SAS Inc., Cary, NC) was used.