Epidemiologic Study of Intensive Care Unit Admission in South Korea: A Nationwide Population-Based Cohort Study from 2010 to 2019

We aimed to investigate the trends of intensive care unit (ICU) admissions in South Korea from 2010 to 2019. We included all adult patients (≥20 years old) who were admitted to the ICU during hospitalization from 2010 to 2019 in South Korea. There were 3,517,423 ICU admissions of 2,461,848 adult patients. Of the ICU admission cases, 66.8% (2,347,976/3,517,423) were surgery-associated admissions, and the rate of in-hospital mortality after ICU admission was 12.0% (422,155 patients). The most common diagnoses were diseases of the circulatory system (36.8%) and pneumonia (4%). The 30-day, 90-day, and 1-year mortality rates were 16.0%, 23.6%, and 33.3% in 2010, and these values slightly decreased by 2019 to 14.7%, 22.1%, and 31.7%, respectively. The proportions of continuous renal replacement therapy (CRRT) use and extracorporeal membrane oxygenation (ECMO) support were 2.0% and 0.3% in 2010, and these values gradually increased by 2019 to 4.7% and 0.8%, respectively. Although the age and cost of hospitalization among critically ill patients who were admitted to the ICU increased from 2010 to 2019, the mortality rate decreased slightly. Moreover, the proportions of ECMO support and CRRT use had increased in our South Korean cohort.


Introduction
The intensive care unit (ICU) is designed to care for critically ill patients who require more support and attention than is available in the general ward [1]. The first intensive care unit (ICU) was established in the late 1950s, and since then, critical care medicine has improved [2,3]. Currently, the ICU plays a critical role in monitoring critically ill patients and providing interventions and organ support [4].
Although critical care medicine and ICUs have a 60-year-long history [5], there are not enough epidemiologic studies on the trends of ICU admission based on big data. Most epidemiological studies have analyzed the trends of surgical or neurological ICU admissions [6,7]. Garland [9]. However, the circumstances of ICU admission differ among countries, depending on the availability of resources for organ support and adequate staffing [10]. In South Korea, the National Health Insurance Service (NHIS) provides nationwide registration data for medical research, including treatment information related to ICU admissions. Thus, using data from the NHIS database, we examined the trends of ICU admission in South Korea from 2010 to 2019. This time frame (2010-2019) was chosen because many advances have been made in critical care medicine in South Korea since 2010 [11].

Ethical Statement
This population-based cohort study complied with the Strengthening the Reporting of Observational Studies in Epidemiology guidelines [12]. The study protocol was approved by the Institutional Review Board (X-2102-666-904) and the Big Data Center of the NHIS (NHIS-2021-1-620). The requirement for informed consent was waived because the data analyses were performed retrospectively using anonymized data derived from the South Korean NHIS database.

Data Source and Study Population
As a single public health insurance system, the NHIS contains and manages the data on disease diagnoses and prescriptions of procedures and/or drugs. In South Korea, physicians (from all outpatient clinics and hospitals) must register all prescription information on procedures, medications, and disease diagnoses in the NHIS database to receive treatment costs from the government. Diseases are registered using the International Classification of Diseases External 10th Revision (ICD-10 codes). Moreover, the NHIS database contains demographic and socio-economic status-related information of all the patients in South Korea.
We included all adult patients (≥20 years) who were admitted to the ICU during hospitalization from 2010 to 2019 in South Korea. The prescription code of ICU admission during hospitalization, available in the NHIS, was used for data extraction.

Data Collection
The following demographic and socioeconomic data were collected: age, sex, employment status, national household income level, and residence at hospital admission. The NHIS contains the patients' household income level that is used to determine insurance premiums in the year, and approximately 67% of the medical expenses are subsidized by the government [13]. However, individuals from low-income households are enrolled in the Medical Aid program; in this program, the government covers nearly all medical expenses to minimize the financial burden of medical costs. The patients were divided into five groups using quartile ratios (Q1 to Q4 groups and Medical Aid program group). Residence was classified into urban (Seoul and other metropolitan cities) and rural residence (all other areas). The lengths of hospital stay (days) and ICU stay were recorded. The admitting departments were classified into internal medicine [IM] and non-IM. We also reported whether the patients were admitted to the ICU through the emergency room (ER). The patients who underwent surgery during hospitalization were considered to have surgery-associated hospital admissions. The hospitals in which the patients were admitted were classified into three groups: tertiary general hospitals, general hospitals, and other hospitals. To determine the comorbid status of the patients, the Charlson comorbidity index (CCI) was calculated using the ICD-10 codes (Table S1). Data on the use of mechanical ventilatory support, extracorporeal membrane oxygenation (ECMO) sup-port, and continuous renal replacement therapy (CRRT) during ICU stay were collected. The follow-up events were classified into four groups: (1) same-hospital follow-up, (2) transfer to a long-term facility care center, (3) death during hospitalization, and (4) discharge and other outpatient clinic follow-up. The dates of death during hospitalization and hospital discharge were also collected. The total cost of hospitalization was collected (in United States Dollar, USD). The main diagnosis at ICU admission was identified using ICD-10 codes. The main diagnosis of all the patients was determined by the NHIS after hospital discharge or death as the disease that required intensive treatment or examination during hospitalization.

Study Objectives
First, the trends of age, total cost, mortality (30-day, 90-day, and 1-year), ECMO support, and CRRT use were examined from 2010 to 2019. Second, we investigated the factors associated with in-hospital mortality and 1-year mortality among all patients admitted to the ICU.

Statistical Analyses
The clinicopathological characteristics of the patients are presented as mean values with standard deviations (SDs) for continuous variables and as numbers with percentages for categorical variables. Multivariable logistic regression modeling was used to determine which factors were associated with in-hospital mortality among patients admitted to the ICU. All covariates were included in the multivariable model for adjustment, and the results were presented as adjusted odds ratios (aORs) with 95% confidence intervals (CIs). The Hosmer-Lemeshow statistic test was used to confirm the goodness of fit of the model. Moreover, we also fitted a multivariable Cox regression model for 1-year mortality among patients admitted to the ICU. The results were presented as adjusted hazard ratios (aHRs) with 95% CIs, and log/log plots were used to confirm that the central assumptions of Cox proportional hazard models were satisfied. There was no multicollinearity between the variables with the criterion of variance inflation factors < 2.0. All statistical analyses were performed using IBM SPSS Statistics for Windows (version 25.0; IBM Corp., Armonk, NY, USA), and statistical significance was set at p < 0.05.   Table 2 lists the main diagnoses at ICU admission. The most common diagnoses were diseases of the circulatory system (I00-I99, 36.8%), followed by neoplasm (C00-D49, 14.1%); injury, poisoning, and certain other consequences of external causes (S00-T88, 10.8%); and diseases of the respiratory system (J00-J99, 10.6%). Table S2 lists the 18 common specific diseases in the main diagnoses at ICU admission. The most common diagnosis was pneumonia (J189, 4%), followed by cerebral infarction (I639, 2.8%), unstable angina (I200, 2.8%), traumatic subdural hemorrhage (S065, 2.4%), acute myocardial infarction (I219, 2.3%), and sepsis (A419, 1.9%).

Trends of ICU Admission
The mean values of age and total cost were 65.

Trends of ICU Admission
The mean values of age and total cost were 65.

Discussion
In this population-based cohort study, we determined the trends of ICU admission in South Korea from 2010 to 2019. The proportion of surgery-associated ICU admissions was 66.8%, with diseases of the circulatory system being the most common main diagnoses. Among specific diseases, pneumonia was the most common cause of ICU admission. The mean age of the patients admitted to the ICU had increased over the years. In addition, although the total hospital cost at ICU admission increased, the 30-day, 90-day, and 1year mortality rates had decreased slightly from 2010 to 2019. The proportion of patients who received CRRT or ECMO had also increased. Our results are different from those of previous studies [6][7][8][9] because our survival analysis included all the patients admitted to the ICU.
We examined the data from 2010 to 2019 to determine the trends of ICU admission in South Korea in this study. In South Korea, on 12 January 2010, a public hearing was held at the National Assembly, where the Korean Society of Critical Care Medicine (KSCCM) expressed the need for intensivists in ICUs to opinion leaders in the legislation, major media outlets, and policy makers [11]. Then, a training system for qualified intensivists was established in South Korea, and there have been advances in this system in the last 10 years. Therefore, it was important to determine the trends of intensive care and hospitali-zation during this period in South Korea.
Interestingly, unemployment and low household income level at ICU admission were associated with a high risk of in-hospital mortality after ICU admission. In the Unit-ed States, low household income was associated with a high risk of in-hospital mortality among patients with sepsis [14]. In our study, in-hospital mortality after ICU admission was high in the Medical Aid program group. Considering that employment reflects the functional status of patients [15], it was not unusual for us to find that unemployment was associated with poor survival outcomes after ICU admission.
Diseases of the circulatory system were the most common diagnoses at ICU admis-sion. The role of the cardiovascular ICU, which is needed for special systemic manage-ment of patients with severe cardiovascular diseases, has recently been emphasized [16]. Moreover, critical care after ICU admission for cardiac surgery has been highlighted as an important factor in recent critical care literature [17].
Regarding specific diseases, pneumonia was the most common disease at ICU ad-mission. Both community-acquired and nosocomial pneumonia are common diseases in ICU-admitted patients that may affect the in-hospital mortality of critically ill patients [18,19]. Considering that we focused on data from patients admitted to the ICU until 2019, the prevalence of pneumonia in the South Korean ICU might have increased due to the coronavirus disease pandemic [20].
We found that the proportion of patients receiving both CRRT and ECMO increased from 2010 to 2019. Other epidemiological studies have also reported that the prevalence of CRRT or ECMO has increased in South Korea [21,22]. Moreover, the increase in CRRT or ECMO might increase the total cost of hospitalization at ICU admission, considering that CRRT and ECMO are relatively expensive procedures. This finding is important because the aging population is susceptible to multiorgan dysfunction, eventually requiring CRRT or ECMO, and the use of CRRT or ECMO will continue to increase in the future given the increase in the aging population.
This study had several limitations. First, some important data, such as body mass index or alcohol consumption history, were not included in this study because of the lack of information in the NHIS database. Second, we did not adjust for disease severity in patients admitted to the ICU with a critical illness, namely the acute physiology and chronic health evaluation II score or simplified acute physiology score II were not used for adjustment. Lastly, the generalizability of the results of this study might be limited because the ICU environment and policies of critical care in other countries are different from those in South Korea.

Conclusions
This study showed that although the age of and cost of hospitalization for critically ill patients who were admitted to the ICU increased from 2010 to 2019, the mortality rate decreased slightly. Diseases of the circulatory system (I00-I99) were the most common main diagnoses, and pneumonia was the most common specific disease at ICU admission. Moreover, the proportions of the use of ECMO and CRRT had increased in South Korea. This study on the recent trends of ICU treatment can help to predict future changes in critical care medicine in South Korea. In addition, our study provides an insight into the aspects of ICU care that need further research and improvement in order to reduce ICU-related mortality in the Korean patient population.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/ijerph20010081/s1, Table S1: The ICD-10 codes used to compute the Charlson comorbidity index, Table S2: 18 common specific diseases in the main diagnoses at ICU admission.

Informed Consent Statement:
The requirement for informed consent was waived because the data analyses were performed retrospectively using anonymized data derived from the South Korean NHIS database.