We used the NEDIS database from 2017 to 2019. In brief, the NEDIS database was established in 2003 on the basis of Article 15 of the Emergency Medical Service Act to evaluate the quality of emergency medical services and provide evidence for the development of a national emergency care plan. Since 2016, over 95% of EDs in Korea have participated in the NEDIS database and transmitted information about visiting patient. The annual number of EDs participating in the NEDIS database was 413 out of 416 in 2017 (99.3%), 399 out of 401 in 2018 (99.5%), and 401 out of 402 in 2019 (99.8%). The NEDIS database design and variables have been described in detail elsewhere [10,11].

We included ED visits by patients over the age of 20. The initial dataset from January 2017 to December 2019 contained a total of 27,410,510 discharge records of ED visits. Visits with missing age (324 visits) and sex (0 visits) information and visits for trauma treatment or unidentified reasons (11,651,209 visits) were sequentially excluded. Cancer related ED visits were defined as cases recorded from C00 to C96 in the primary or secondary ED diagnosis using the Korean Classification of Diseases, 7th revision (KCD-7), which is a modified version of the International Classification of Diseases, the 10th revision (ICD-10). In Korea, patients being treated for cancer should be enrolled in electronic medical records with an accurate cancer diagnosis to receive special benefits from the National Health Insurance (NHI). As the study aimed to investigate ED visits of cancer patients who were being treated, cases that were not coded as cancer in the ED but were subsequently coded as cancer after hospitalization were excluded. After the exclusion of 3,513,306 visits by cancer patients aged under 19, the final number of study subjects included 667,935 cancer-related visits and 11,445,250 non-cancer-related visits out of a total of 12,113,185 visits (Fig. 1).

The data contained demographic characteristics such as age, sex, insurance status, and information about ED utilization, including ED type, ED region, hospital service level, arrival time, and patient transportation. Clinical variables of initial triage scale, vital signs, primary cancer type, disposition, ED diagnosis, and diagnosis after hospitalization were also included. Initial triage was designated according to the Korean Triage and Acuity Scale (KTAS), which prioritizes patients according to the five ordinal scales reflecting both clinical severity and acuity (resuscitation, emergent, urgent, less urgent, non-urgent). For service level of hospital (certified tertiary hospital, general hospital, hospital), we applied the classification of the Medical Law of Korea (https://www.law.go.kr/laws/medical law). A “hospital” is a medical institution with 30 or more beds and provides treatment mainly for inpatients, and a “general hospital” has more than 100 beds with some requirements for more than or less than 300 beds. “Certified tertiary hospital” specializes in medical treatment with high difficulty for severe diseases, which can be designated by the Ministry of Health and Welfare of Korea. The variables of insurance status (NHI, Medicaid, uninsured), type of ED (regional emergency center, local emergency center, local emergency institution), region of ED (metropolitan city, province), mode of arrival (ambulance, non-ambulance), time of arrival (9 am–6 pm, 6 pm–9 am), LOS in the ED (under 2 hours, 2–3 hours, 4–7 hours, more than 8 hours), disposition (discharged to home, transferred to another hospital, admitted, death in the ED), and admission destination (general ward, intensive care unit) were categorized.

We compared the demographic, clinical, and disposition characteristics between cancer-related and non-cancer-related visits. Among the cancer-related visits, we assessed differences according to the hospital service level. Comparisons between the groups were performed using chi-squared test on categorical variables; continuous variables were categorized as described above. We also calculated frequency and percentage of the most common cancer type and primary diagnosis for the cancer-related visits.

Multivariate analysis was conducted to investigate the independent predictors associated with the LOS in ED and hospital admission. For hospital admission, potential predictors including age, sex, insurance status, ED region, hospital service level, time of arrival, mode of arrival, KTAS classification, and LOS. Disposition status were tested with univariate logistic regression analysis and entered to multivariate model with forward stepwise entry of statistically significant parameters. A similar analysis was performed with prolonged LOS. The prolonged LOS defined as LOS in ED more than 8 hours based on existing literature [12,13].

All tests were two-tailed, and results with p-values < 0.05 were considered statistically significant. All data preparation and statistical analyses were conducted using SAS ver. 9.4 (SAS Institute Inc., Cary, NC).

ED visits by cancer patients comprised 5.5% (667,935 visits) of total ED visits from 2017 to 2019. The number of ED visits would not have been underestimated because visits for trauma treatment or unidentified reasons were all excluded in non-cancer-related visits, and the diagnostic code for cancer in the ED was perceived as relatively accurate to provide patients with a special benefit from NHI. Male cancer patients visited the ED more frequently than female cancer patients (57.9% vs. 42%) (Table 1). Cancer patients were more likely than non-cancer patients to visit larger hospitals, such as EDs in certified tertiary hospitals (67.5% vs. 30.4%). Cancer-related ED visits were more concentrated during the daytime than in the evening, and 29% of cancer patients stayed in the ED longer than eight hours. Cancer patients stayed in the ED for an average of 8.7 hours (±17 hours) and while non-cancer patients stayed for 3.5 hours (±8.3 hours). When the average LOS in the ED was multiplied by the number of visits, cancer-related visits comprised 12.7% of total ED visits by time. Approximately half of ED visits by cancer patients led to hospitalization (50.4%).

The most common cancer-related ED visits were for patients with lung cancer (14.1%) and liver cancer (12.2%) (Table 2). ED visits associated with colon, stomach, breast, and pancreatic cancers followed. Patients with cancer exhibited a range of primary or secondary ED diagnoses (Table 3), including fever, pneumonia, abdominal and pelvic pain, other gastroenteritis and colitis, paralytic ileus, and intestinal obstruction.

The characteristics of cancer-related visits were displayed by a subgroup of hospital levels in Table 3 because the rate of cancer-related ED visits varied greatly by hospital service levels in Table 1. After further exclusion of cases missing the LOS in the ED and cases in Sejong City (administrative capital) in consideration of the specificity, a total of 667,864 visits were analyzed. A higher proportion of male visits is associated with lower hospital service levels. Higher hospital service levels are associated with more daytime visits, longer ED stays, and higher KTAS levels. The proportion of visits that lasted longer than eight hours in certified tertiary hospitals was 36.3%, compared to 14.3% in general hospitals. However, the proportion of admitted cases in certified tertiary hospital (45.2%) was slightly lower than in general hospitals (48.0%).

According to multivariate analyses, cancer patients stayed longer in EDs located in metropolitan cities (adjusted odds ratio [aOR], 1.44; 95% confidence interval [95% CI], 1.422 to 1.460) than in provinces, and cancer patients without insurance were hospitalized less (aOR, 0.80; 95% CI, 0.708 to 0.892) than NHI patients (Table 4). Hospital service level was the most significant factor associated with the LOS of cancer patients in the ED. The aOR for a prolonged LOS was 126.34 (95% CI, 89.446 to 178.448) in the ED of certified tertiary hospitals compared to hospitals. Patients who visited the ED during the daytime stayed for a shorter amount of time in the ED and were admitted more than those who visited the ED in the evening. The aOR for admission in certified tertiary hospitals was lower than that of general hospitals and hospitals, although a more critical clinical status according to the KTAS classification was significantly associated with a higher admission rate (Table 4). Nevertheless, the longer patients stayed in the ED, the more likely they were to be admitted.