Facility type as the primary determinant of hospitalization in pediatric foreign body ingestion: a nationwide emergency department-based analysis

Article information

Pediatr Emerg Med J. 2026;13(3):93-104

Publication date (electronic) : 2026 June 9

doi : https://doi.org/10.22470/pemj.2026.01585

Yujin Lee¹, Dong-Uk Kim^,2

¹Department of Medicine, Inje University College of Medicine, Busan, Republic of Korea

²Department of Pediatrics, Seoul National University College of Medicine, Seoul, Republic of Korea

Corresponding author: Dong-Uk Kim, Department of Pediatrics, Seoul National University College of Medicine, 101 Daehak-ro, Jongno-gu, Seoul 03080, Republic of Korea Tel: + 82-2-2072-3566; E-mail: donguk.kim90@gmail.com

Received 2026 March 9; Revised 2026 April 2; Accepted 2026 May 20.

Abstract

Introduction

Foreign body (FB) ingestion is a common pediatric presentation in emergency departments (EDs) worldwide (1,2). The majority of cases occur in children younger than 6 years, with a slight male predominance (1,3-5). Recent epidemiologic studies indicate a rise in pediatric FB ingestion, highlighting its growing clinical and public health relevance (4,6). Although most FB ingestion cases resolve spontaneously (1,2,4), certain objects, such as batteries or sharp objects, can result in serious complications such as bowel perforation or infection (7). The time-sensitive nature of these complications has led to the international consensus guidelines that emphasize the importance of timely endoscopic intervention (8). However, the availability of such specialized emergency interventions may vary depending on regional healthcare infrastructure.

South Korea exhibits regional disparities in healthcare resource distribution that may affect the management of time-sensitive conditions. Despite the country’s compact geography, medical resources are disproportionately concentrated in the capital region, which comprises Seoul, Incheon, and Gyeonggi Province and houses 50% of the national population (8,9). Recent studies on avoidable and preventable mortality show that while absolute health outcomes have improved nationwide, relative disparities between the capital and non-capital regions have stagnated or widened (10). This pattern of healthcare centralization mirrors global trends showing reduced access to specialist care and lower healthcare utilization in rural areas (11,12).

In addition, timing for FB-removing interventions requires assessment of various factors including object type, anatomical location, symptoms, and time since ingestion. The timing is divided into 3 categories: emergent (<2 hours from presentation, regardless of nil per os), urgent (<24 hours from presentation), and elective (>24 hours from presentation) (13). For example, esophageal coin ingestion is considered emergent if the patient is symptomatic but urgent if asymptomatic. In contrast, coin ingestion in the stomach or small bowel without any symptoms is assigned to the elective category. This categorization of the timing shows that regional disparities in access to pediatric gastroenterology services and specialized equipment may disproportionately affect clinical outcomes in non-capital regions.

Despite these concerns, there is a lack of national-level evidence addressing regional disparities in pediatric FB ingestion care. International studies have discussed socioeconomic disparities in pediatric FB ingestion management (14-17), but Korean studies have been limited to small, single-center descriptive cohorts (18,19). Larger Korean studies have only focused on risk factors for endoscopic complications (20,21) or specific types of FBs (22). Using Korean nationwide data from the National ED Information System (NEDIS), we aimed to investigate the regional disparities in hospitalization rates and facility utilization patterns among children presenting with FB ingestion.

Methods

Results

1. Characteristics of the pediatric ED visits for FB ingestion

A total of 2,870 pediatric ED visits for FB ingestion were recorded between 2019 and 2023 (Table 1). Meanwhile, the number of patients aged 10–19 years was 463, accounting for 13.9% of the eligible cohort (Appendix 4 [https://doi.org/10.22470/pemj.2026.01585]). The visits involved predominantly male patients (55.6%), with the largest age group being 1–4 years (55.3%). Only 22.6% were categorized as high acuity. Data for KTAS levels are detailed in Appendix 5 (https://doi.org/10.22470/pemj.2026.01585). Consistent with overall low acuity, most visits resulted in discharge (89.7%), with an overall hospitalization rate of 8.1%. More than half of the visits occurred at the local EMCs (51.3%), followed by the regional EMCs (36.1%) and local EMFs (12.5%).

Table 1.

Clinical characteristics of pediatric emergency department visits for foreign body ingestion (N = 2,870)

The study period encompassed the coronavirus disease 2019 pandemic, which may have altered the ED utilization pattern. However, Appendix 6 (https://doi.org/10.22470/pemj.2026.01585) indicates that the regional disparities in clinical disposition remained unchanged despite a minor fluctuation in total volume of the ED visits. Regardless of the pandemic, the non-capital regions consistently exhibited lower hospitalization and higher transfer rates.

2. Geographic distribution of ED visits

To examine the regional disparities, we first assessed the geographic distribution of population-adjusted ED visit rates across Korea (Figure 1 and Appendix 7 [https://doi.org/10.22470/pemj.2026.01585]). Seoul recorded the highest rate at 168.1 per 100,000 children, substantially exceeding all other regions including major metropolitan cities such as Daegu (114.5) and Gwangju (111.1). Even Gyeonggi Province (69.7) and Incheon (53.7) also showed higher rates than most non-capital regions. This capital-centric pattern is clearly illustrated on the national map (Figure 1), with visit rates forming a gradient radiating from Seoul. To explore whether this geographic variation translated into differences in clinical outcomes, we compared hospitalization-related characteristics between the capital and non-capital regions.

Fig. 1.

Regional distribution of pediatric foreign body ingestion: ED visit rate per 100,000 inhabitants. The names of metropolitan cities are marked outside the map using the line, while those of provinces are marked directly on the map. ED: emergency department.

3. Regional disparities in the clinical characteristics and paradoxical hospitalization pattern

Table 2 compares the demographic and clinical characteristics of ED visits originating from the capital (n = 1,864) and non-capital (n = 1,006) regions. Compared with the capital-region visits, the non-capital visits showed a higher percentage of high-acuity (21.6% vs. 24.4%; P <0.001) and more frequently observed vomiting, abdominal pain, and sore throat (See distribution of associated symptoms in Appendix 5). Despite the higher acuity, a larger proportion of the non-capital visits were managed at the local EMFs, compared with the capital-region visits (3.9% vs. 28.6%; P <0.001). Notably, despite the higher clinical acuity, the non-capital visits showed a numerically lower hospitalization rate (8.9% vs. 6.7%) and a higher transfer rate (1.7% vs. 2.7%; all Ps = 0.053). This pattern, combined with the markedly higher proportion of care delivered at the local EMFs in the non-capital regions, suggested that facility type, rather than geography per se, may be the key determinant of hospitalization.

Table 2.

Characteristics according to the region

4. Facility type as the primary determinant of hospitalization

Table 3 presents the regression findings. Model 1 showed capital-region residence, visits to the EMCs, age of 1 year or older, higher acuity, and specific symptoms (vomiting, abdominal pain, and fever) as factors associated with hospitalization.

Table 3.

Univariable and multivariable logistic regression of factors associated with hospitalization

Among the significant factors in Model 1, the capital-region residence (adjusted odds ratio, 1.76; 95% confidence interval, 1.29–2.43), higher acuity (2.96; 2.22–3.94), and the 3 symptoms (vomiting [2.88; 1.71–4.80], abdominal pain [5.69; 3.40–9.52], and fever [3.34; 1.58–6.74]) remained significant in Model 2. In Model 3, the facility type emerged as the strongest predictor. Compared with the local EMFs, hospitalization was markedly more likely at the regional EMCs (adjusted odds ratio, 14.20; 95% confidence interval, 5.19–58.70) and the local EMCs (8.35; 3.03–34.60). In addition, high acuity remained a significant predictor of hospitalization (2.66; 1.98–3.57), as well as vomiting, abdominal pain, fever, and capital-region residence.

Model fit indices indicated that adding facility type improved model performance (Table 3). Model 3 demonstrated better calibration than Model 2 (Hosmer–Lemeshow χ² = 4.14 vs. 24.42, df = 8; P = 0.844 vs. 0.002; the higher P-value, the better fit) and higher explanatory power (McFadden R² = 13.8% vs. 10.9%). Likelihood ratio test confirmed that this improvement was significant (P <0.001).

The interaction between the region and facility type on hospitalization rate is illustrated in Figure 2. Although the overall hospitalization rate was higher in the capital region (8.9% vs. 6.7%; Table 2), the rate of the regional EMCs was paradoxically higher in the non-capital regions. This regional disparity can be explained by the different patterns of visits to local EMCs and EMFs. The visits in the capital region were concentrated in the local EMCs (60.4%), of which hospitalization rate was 8.9%. In contrast, patients in the non-capital regions visited the local EMFs (28.6%) more frequently than those in the capital regions (3.9%). Taken together, these findings indicate that geographic variation in pediatric FB ingestion hospitalization largely reflects differential access to higher-level facilities, rather than intrinsic regional differences in clinical severity or decision-making, with the facility type emerging as the most influential determinant of hospitalization (Table 3).

Fig. 2.

Hospitalization rate by the region and facility type. EMC: emergency medical center, EMF: emergency medical facility.

5. Analysis of the mediating role of the facility type

For this analysis, the facility type was dichotomized into the regional and local EMCs vs. the local EMFs. The capital-region patients had 8.74 times higher odds of visiting an EMC (Path A in Figure 3; Appendix 8 [https://doi.org/10.22470/pemj.2026.01585]). Moreover, receiving care at an EMC was associated with 13.7-fold increased odds of hospitalization (Path B; Appendix 9 [https://doi.org/10.22470/pemj.2026.01585]). After adjusting for the facility type, the direct association between the capital-region residence and hospitalization was attenuated and no longer significant (Path C; Appendix 9). Overall, the capital-region residence was associated with a 3.47% increase in hospitalization rate, of which 1.50% was flowing through the facility type (indirect effect), representing 43.0% of the total effect (Table 4). These findings suggest that nearly half of the regional disparities in the hospitalization rates can be attributed to differences in the facility use patterns (Appendix 10 [https://doi.org/10.22470/pemj.2026.01585]).

Fig. 3.

Mediation analysis of the association between region and hospitalization through the facility type. EMC: emergency medical center, EMF: emergency medical facility, aOR: adjusted odds ratio, CI: confidence interval.

Table 4.

Mediation analysis of the association between capital-region residence and hospitalization through facility type

Discussion

This study shows the paradoxical pattern of pediatric FB ingestion hospitalization in Korea. The number of ED visits due to FB ingestion is concentrated in the capital regions, especially in Seoul. The acuity in the regions was lower than that in the non-capital regions. In contrast, the non-capital regions showed a higher acuity and greater reliance on lower-level facilities, yet paradoxically showed a lower hospitalization rate and higher transfer rate than capital regions. The regression analysis demonstrated that the facility type was the strongest predictor of hospitalization and that the regional effect was attenuated when the facility access was considered.

Our findings reinforce established global epidemiological trends regarding pediatric FB ingestion. Consistent with studies from Europe (1) and the United States (3,4), more than half of ED visits in our cohort were male or younger than 4 years (Table 1). Clinically, most presentations were low-acuity or resulted in discharge. Compared to Korean pediatric ED visits for various reasons, the proportion with KTAS 4-5 level was higher (76.0% vs. 60.2%) in FB ingestion cases (25). This suggests that FB ingestion presentations were less serious compared to other ED chief complaints, aligning with the consensus that most pediatric ingestions resolve spontaneously (2). However, the remaining cases requiring hospitalization or transfer presented with higher acuity, underscoring the critical role of appropriate facility access.

A key finding of this study was the paradox between acuity and facility utilization in the non-capital regions, as detailed in the section “Regional disparities in the clinical characteristics and paradoxical hospitalization pattern.” This paradox could be explained by the natural history of pediatric FB ingestion. Since 80%–90% of FB ingestion cases resolve spontaneously without endoscopic intervention or surgery (2), the initial presentation to nearby local EMFs in the non-capital regions may be appropriate for the majority. Although the other 10%–20% require interventions, the local EMFs often lack the specialized equipment and personnel to provide definitive care. This structural limitation likely explains both the lower hospitalization and higher transfer rates in the non-capital regions (Table 2) where clinicians at the local EMFs are often left with transfer as the only viable option for the minority requiring definitive interventions.

The independent association of hospitalization with some symptoms or high acuity aligns with previous studies noting frequent gastrointestinal symptoms in hospitalized patients (19). However, even after adjusting for these strong clinical indicators, the facility type remained vital as a determinant of hospitalization, particularly the regional EMCs (Table 3). This finding may be explained by referral bias, whereby higher-volume hospitals treat more severe cases (26,27).

The significant improvement in model fit (McFadden R2 13.8% vs. 10.9%) after adjusting the facility type also suggests that hospitalization is not solely decided by patient-level factors but also by each ED’s capability. The regional disparities in hospitalization are not based on geographic limitations but a reflection of the facility effect. The finding that the non-capital regional EMCs showed a higher hospitalization rate (13%) than capital (11%) suggests that high-level facilities are utilized regardless of the region. Therefore, the lower hospitalization rate in the non-capital areas could stem from the disproportionate reliance on the local EMFs, given that EMFs showed the lowest hospitalization rate and were more prevalent in non-capital regions than in the capital region. The possibility that 43% of children might flow through the type of facility may be driven by a fundamental imbalance in the distribution of emergency healthcare infrastructure in Korea. Capital region’s emergency facilities show a more balanced composition (EMC, 59.0% vs. EMF, 41.0%) compared to such facilities in the non-capital region (EMC, 35.6% vs. EMF, 64.4%) (28). As more regional or local EMCs are concentrated in the capital region, the likelihood of hospitalization after FB ingestion may be influenced more by the structural composition of the region’s healthcare supply than by the clinical severity.

These structural disparities warrant attention regarding healthcare financing. Korea’s National Health Insurance Service, which covers the entire population in the country, bears the cost of variations in hospitalization. Given that hospitalization substantially raises costs in pediatric FB ingestion management (5), hospitalization decisions driven by the facility type rather than clinical need may represent inefficient allocation of healthcare resources. This pattern is consistent with another research showing the uneven distribution of emergency care capacity in Korea. For high-acuity conditions, regions without EMCs showed a higher risk of 30-day mortality compared to regions with EMCs (29).

Prior U.S.-based research has primarily focused on socioeconomic associations with FB ingestion. Some studies have shown associations between lower-income households and higher FB ingestion rates (14,15). However, the opposite result was also shown. Button battery ingestion occurred in the U.S. households with a higher median income (16). Saw-Aung et al. (17) identified relationships between race and FB ingestion frequency. In contrast, our study highlights healthcare infrastructure as the key determinant of hospitalizing children with FB ingestion in Korea.

This study has several limitations inherent to its retrospective study design and the use of administrative data. First, while we identified a strong association between the facility type and hospitalization, the design precludes the establishment of causality. Second, given the nature of a general emergency surveillance system, the NEDIS database did not provide clinical details, such as the type or size of FBs, spontaneous passage or endoscopic removal, or post-discharge complications. Third, most importantly, we could not measure potential confounders of hospitalization decision, such as physicians’ subspecialty, family preferences on selecting facility types, and institutional protocols or capabilities. In particular, the on-site availability of pediatric gastroenterologists or surgeons may mediate the observed facility effect, as hospitalization decisions are ultimately made by such specialists. Fourth, the dichotomous regional classification may not reflect the clinical heterogeneity of non-capital, metropolitan cities, such as Busan. Lastly, this study did not use multilevel modeling to account for clustering at the institutional level. Despite the limitations, to our knowledge, this study is the first national-level analysis investigating the impact of regional disparities on pediatric FB ingestion-related hospitalization, establishing a foundation for future studies on geographic inequities in pediatric emergency care.

In summary, this nationwide study demonstrates facility-driven disparities in pediatric FB ingestion hospitalization in Korea. Patients from the non-capital regions presented with higher acuity but heavily relied on lower-level emergency facilities, and were less frequently hospitalized, compared to those from the capital regions. In the regression model, beyond clinical severity alone, the availability of high-level facilities may be a primary determinant of hospitalization. Mediation analysis suggests that unequal distribution of EMCs depending on the geographic region may contribute to the regional disparity in hospitalization. Addressing these disparities may require a tiered regional strategy. For example, we recommend establishing hub hospitals, such as regional EMCs or selected local EMCs, in each non-capital region so that pediatric endoscopic and surgical management is available on a 24/7 basis. Concurrently, local EMFs should function as first-contact facilities supported by streamlined referral pathways and standardized disposition protocols to ensure that the minority requiring definitive intervention receives timely definitive care regardless of geographic location.

Article information Continued