Introduction

Patient visits to the emergency department (ED) have been increasing worldwide for several decades [1,2]. Although the precise reasons are not known, changes in transportation, increased leisure time, and improved access to medical service could be factors. Pediatric patients account for a significant portion of ED visits (31.2% in Korea and 20.5% in the US) [3,4]. Many ED patients visit EDs for illnesses or injuries that do not require emergency treatment, or for conditions that are treatable or preventable by primary care [5,6], and previous studies reported that 58% to 82% of all pediatric patients visited the ED with non-urgent conditions [7,8]. Possible reasons might be the lack of appropriate alternative medical facilities, convenience of ED access, or overestimation of the severity of children’s symptoms by parents [9,10].

The increase in absolute number of ED visits and proportion of non-urgent visits are the main factors affecting ED overcrowding [11]. Overcrowding is problematic in that it can impede appropriate and timely treatment due to diversion of limited resources, resulting in poor prognoses and negatively affecting working staff and patient satisfaction [12]. To resolve ED overcrowding, it is important to first identify the characteristics of ED visits, and based on this, intervention strategies could be explored.

In Korea, patients can use ED services directly whenever necessary without a prior primary clinic visit, regardless of the acuity or severity of illness. In 2016, 539 EDs were available around the country. With regard to the characteristics of pediatric ED visits in Korea, Kwak et al. [3] reported an analysis of nationwide data from the National Emergency Department Information System (NEDIS) of Korea. That study primarily investigated the clinical characteristics of patients at the time of the ED visit. However, in addition to patients’ clinical characteristics, demographic factors also affect use of ED services. Therefore, a comprehensive analysis of pediatric ED patients investigating both clinical and demographic characteristics can provide baseline data for establishing desirable policies for pediatric EDs. In this study, we investigated characteristics of pediatric ED patients and their families, and explored characteristics related to hospital admission by analyzing nationwide panel data in Korea.

Methods

After approval of this study by the ethics committee of Inje Univsersity Ilsan Paik Hospital (Goyang, Korea), we analyzed Korea Health Panel Study (KHPS) data (version 1.2.2) for a six-year period (2008–2013). Due to the retrospective nature of our study, the requirement for informed consent was waived. In addition, we obtained KHPS data that were fully anonymized before analysis.

Survey data source

The KHPS is an official statistical investigation and has been conducted annually since 2008 under the supervision of the Korea Institute for Health and Social Affairs and the National Health Insurance Service. The KHPS aims to acquire baseline data for establishing and improving health care policies and health insurance policies by examining data related to medical service use, medical expenditures, and insurance coverage. The sampling frame was set to 90% of the national population surveyed in the 2005 Population Census of Korea, and a two-stage probability proportionate and stratified cluster sampling method was used. First, the population was stratified according to geographic area using household registries (16 metropolitan cities and provinces, and two towns [dongs and eups/myeons]); this yielded a total of 237,165 clusters. From the whole population clusters, 350 sample clusters were extracted, and then sample households were extracted from those clusters. Finally, family members from these sample households became the Korean Health Panel [13]. At the start, about 8,000 households across the nation were sampled, and the survey was conducted once or twice a year using both self-reporting questionnaires and in-person interviews with trained investigators. Response rates during each year of the survey are shown in Table 1. Receipts for medical expenses and prescriptions were used as supporting information about use of medical services by household members. Most of the children’s data were collected by their parents, and most surveys were conducted within one year of each ED visit. However, the exact time gap between the ED visit and survey completion was not fixed.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. KHPS survey response rates.

https://doi.org/10.1371/journal.pone.0197929.t001

Results

Overall characteristics

A total of 3,160 pediatric ED visits occurred during the study period. Table 2 shows that the proportion of male children (57.5%) was higher than that of females, and of the age groups, those aged 0–5 years visited EDs the most (47.7%). For residence region, province ranked highest, followed by metropolitan city and capital. For household income, the fourth quintile produced the most visits, followed by the third and first quintiles. Among all children, 87.4% were covered by private health insurance, and 40.5% visited the ED more than twice a year. Transportation via ambulance accounted for only 5.2% of visits, and disease as the reason for the ED visit was more than twice as common as injury/poisoning as the reason. Among all the medical services provided in the ED, examination without medication or surgical intervention was most common (88.3%). In terms of size, visits to general hospitals were most common (53.2%). Public hospitals accounted for only 7.8% of visits. Only 14.3% of children required hospital admission or transfer to another hospital as a result of the ED visit.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Case characteristics during study period.

https://doi.org/10.1371/journal.pone.0197929.t002

Discharge group versus admission group

Table 3 shows that male children were more likely than females to require hospital admission, and children aged 6–11 years formed a smaller proportion of admissions than did the other age groups. The proportion of admissions was larger in the first and second household income quintiles than in the fourth and fifth quintiles. Medicaid beneficiaries were more likely to be admitted than National Health Insurance (NHI) beneficiaries, as were those not covered by private insurance. ED visits via ambulance and due to disease formed larger proportions of the admission group than those via self-transportation and due to injury/poisoning, respectively. Patients requiring a surgical operation were more likely to be admitted than those receiving other ED services.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Characteristics of discharge group and transfer/admission group.

https://doi.org/10.1371/journal.pone.0197929.t003

Factors associated with hospital admission

Table 4 shows that variables significantly associated with hospital admission included sex, age, region, private insurance coverage, frequency of ED visits, transport method, reason for visit, and medical services provided. Females exhibited an odds ratio (OR) of 0.73 compared with males; children aged 6–11 years showed an OR of 0.68 compared with those aged 0–5. Residents in provinces had an OR of 0.69 compared with those in the capital, and children in the fifth quintile of household income had an OR of 0.51 compared with the first quintile. Privately insured children had an OR of 0.49 compared to those lacking private insurance, and children with multiple annual ED visits had an OR of 1.33 compared to those with a single annual ED visit. The OR for ambulance transport was 2.98 compared with self-transport, and patients visiting the ED due to disease had an OR of 1.82 compared with those visiting dues to injury/poisoning. Patients who underwent a surgical operation had higher odds than those who received only a diagnostic examination (OR = 5.4).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Factors associated with transfer/admission and their odds ratio.

https://doi.org/10.1371/journal.pone.0197929.t004

Discussion

According to Kubicek et al. [10], demographic factors such as medical insurance coverage, type of insurance coverage, and low household income were closely related to ED visit frequency and severity. Although previous studies in the US evaluated individual and household demographic characteristics of pediatric ED patients [14], previous reports on pediatric ED patients in Korea have focused mainly on clinical characteristics [3,15]. To the best of our knowledge, our study is the first to explore both clinical and demographic characteristics with regard to ED use among the Korean pediatric population.

In terms of sex and age group, our study found that males made more ED visits than females and younger children (age 0–5) made more visits than older children. These findings are similar to those of Kwak et al. [3] in Korea and Ben-Isaac et al. [14] in the US. Korean childrenin the two highest quintiles of household income (49.2%) visited the ED more often than those in the two lowest quintiles (24.7%), perhaps due to the relatively high cost of ED services. This trend was also observed in the US, where middle- and upper- income (≥ 200% of federal poverty level, 56.1%) children made more ED visits than children from poor households (≤ 124% of federal poverty level, 27.5%) [14].

In our study, the rate of ambulance transport (5.2%) was similar to that reported by Kwak et al. [3] (5.3%), and our findings for visit reason (disease being 2.1 times more common than injury/poisoning) closely matched those of Kwak et al. [3], who reported disease as being 2.5 times more common. We conducted post-hoc subgroup analysis to explore the difference in visit reason according to sex, and found a significant difference (p<0.001). ED visits for injury/poisoning were 1.8 times greater in males than in females, whereas ED visits for disease were only 1.2 times greater in males than in females. This discrepancy might be due to higher levels of curiosity and risk-taking among boys [16].

The odds of hospital admission were lower in females than males (OR = 0.73); this was similar to the report of Kwak et al. [3], in which the odds for males were 1.11 times higher than odds for females. The odds of hospital admission for children living in the capital were lower than for those living in provinces. We speculate that relatively high accessibility to the ED enabled capital-dwelling patients to visit the ED more frequently even if their conditions did not necessitate admission. Children in the highest quintile of household income and children covered by private insurance had lower odds for hospital admission than those in the lowest income quintile and those lacking private insurance, respectively. Again, this might reflect the ability of children with high household income or private insurance to visit the ED more frequently, even for conditions not of requiring admission. Visits for disease had higher odds of hospital admission than visits for injury/poisoning, as was previously reported by Kwak et al. [3].

To establish policies for desirable changes to the emergency medical service system, comprehensive analysis of the current state of the system is required. In that respect, our study results can be used as baseline data for future research. For example, based on our finding that children with injury/poisoning had lower odds of hospital admission than children with disease, additional studies regarding children visiting the ED after injury/poisoning could be conducted to investigate whether it would be possible to lower admissions due to injury/poisoning further still. To do this, descriptive analysis of injured children with regard to more detailed injury characteristics and injury related outcomes would be required. In addition, interventional studies investigating the efficacy of strategies to reduce non-urgent ED visits (for example, public relations about first aid for frequent injuries) could be conducted.

One limitation of this study is the possibility of recall bias because the data were collected retrospectively. However, to minimize this bias, data were collected by trained investigators, and household accounts of health service usage were cross-referenced to receipts for medical costs for each panel household. Another limitation is that we only examined the association between characteristics of pediatric ED patients and hospital admission after ED treatment. We did not investigate criteria for distinguishing urgent versus non-urgent emergencies other than hospital admission. As a result, detailed analysis of patients who were admitted to the emergency room with regard to urgency was not possible. Lastly, given the retrospective study design, some possible confounding factors affecting hospital admission might have been missed.

In conclusion, the only medical service provided to the majority of pediatric ED patients was diagnostic examination, and only 14.3% of ED visits resulted in hospital admission. Male sex, residence in the capital, private insurance coverage, and disease as a reason for the ED visit were closely associated with hospital admission. Based on the results of this study, prospective interventional studies might be required to establish policies regarding appropriate and effective pediatric ED service in the future.