Spatial analysis of underserved areas for emergency medical services assessed out-of-hospital cardiac arrest in Jeju

JongYeon Kang; SeoYoung Ko

doi:10.22730/jmls.2025.10.01.01

J Med Life Sci > Volume 23(1); 2026 > Article

Kang and Ko: Spatial analysis of underserved areas for emergency medical services assessed out-of-hospital cardiac arrest in Jeju

Original Article

Journal of Medicine and Life Science 2026;23(1):17-23.

Published online: October 1, 2025

DOI: https://doi.org/10.22730/jmls.2025.10.01.01

Spatial analysis of underserved areas for emergency medical services assessed out-of-hospital cardiac arrest in Jeju

JongYeon Kang¹

, SeoYoung Ko²

¹Department of Emergency Medicine, Jeju National University Hospital, Jeju, Republic of Korea

²Department of Emergency Medicine, Jeju National University Hospital, Jeju National University College of Medicine, Jeju, Republic of Korea

Correspondence to SeoYoung Ko Department of Emergency Medicine, Jeju National University Hospital, Jeju National University College of Medicine, 15 Aran 13-gil, Jeju 63241, Republic of Korea
Tel: 82-64-717-1904 Fax: 82-64-717-1907 E-mail: koseoyoung84@gmail.com

Received July 16, 2025 Revised September 10, 2025 Accepted October 1, 2025

This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0/) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.

Abstract

Geographic information systems are useful tools for analyzing and explaining the changing spatial structures of medical services. This study identified underserved areas for out-ofhospital cardiac arrest (OHCA) using a geospatial methodology and provided a visual map of the spatial imbalances of OHCA. This study used geocoding of OHCA locations, emergency medical service (EMS) incidents, and administrative districts to demonstrate a technique for identifying underserved areas for EMS-assessed OHCA. We created a 1 km hexa-grid across Jeju Island and calculated the OHCA ratio (OHCA cases/total EMS incidents) for each grid. Underserved areas were detected through a hexagrid-based geospatial analysis (Getis-Ord Gi* statistics). A total of 366,155 EMS incidents occurred during the study period. Among these, 6,806 were EMS-assessed OHCA cases. An analysis of 1,903 overlapping grids revealed that the mean distance between grids with OHCA was 28.0±15.8 km. The mean EMS-assessed OHCA ratio (EMS-assessed OHCA cases/total EMS incidents) for each grid was 0.021±0.059. Based on the Getis-Ord Gi* OHCA ratio, 118 hot spot grids (z-score≥1.96) and 19 cold spot grids (z-score≤-1.96) were identified. This study proposed a geospatial methodology to define and identify underserved areas of EMS-assessed OHCA cases using hexa-grid-based hot- and cold-spot analysis.

Key words: Emergency medical services; Geographic information systems; Out-of-hospital cardiac arrest; Spatial analysis

INTRODUCTION

Out-of-hospital cardiac arrest (OHCA) is a time-sensitive emergency encountered by emergency medical services (EMS), along with acute myocardial infarction, acute stroke, and major trauma. For OHCA, the recommended response time is less than 4 minutes for basic life support (BLS) and less than 8 minutes for advanced cardiac life support (ACLS) [1-3]. The longer the response time, the lower the chances of patient survival and a favorable prognosis [4-8].

Geographic information systems (GISs) effectively collect, store, update, adjust, analyze, and present various types of geospatial reference information that exist in the real world using various data models [9-11]. Recently, GISs and related spatial analysis methods have incorporated the concept of space into healthcare, and enabled a spatiotemporal approach. Additionally, geospatial techniques have been introduced as tools to explain and understand the spatial organization of healthcare and to explore ways to improve healthcare delivery and outcomes [12,13]. Such information systems can efficiently support planning and decision-making related to geographical distribution. Unlike other information systems, GISs contain spatial data that can be categorized into positional and attribute data. For example, a road contains position information that indicates its location (latitude, longitude, shape, and coordinates) and attributes (name, length, speed limit, and direction). Therefore, a GIS can identify general patterns of points, lines, and surfaces in the real world and determine their spatial relationships, which cannot be easily derived from existing data [11,14-17]. However, considerable uncertainty remains regarding the geographic approach in the field of OHCA, as most existing studies have reported OHCA statistics without incorporating geospatial information.

This study aimed to apply a GIS approach to detect areas underserved by EMS. We present a geospatial methodology for identifying underserved areas for EMS-assessed OHCA using a geocoding technique and information on EMS-assessed OHCA location. Thus, this study focused on the geospatial analysis of EMS-assessed OHCA demand.

MATERIALS AND METHODS

1. Study approach flow

In this study, the extraction technique for identifying underserved areas of EMS-assessed OHCA involved three steps.

Step 1: preparing spatial data for the predefined area. First, define the analysis area. Second, prepare the administrative district shapefiles containing spatial information. Third, create grid-based shapefiles within the defined geographic units.

Step 2: geocoding of EMS incidents using the text-based address. First, define the OHCA and non-OHCA cases. Second, geocode EMS incident addresses using GIS software.

Step 3: overlay and hot/cold spot analysis in each grid. First, calculate the OHCA ratio (OHCA cases/total EMS incidents) for each grid-based layer. Second, test for global spatial autocorrelation using Moran’s I statistics. Third, identify hot and cold spots as spatial outliers using Getis-Ord Gi* statistics.

2. Study design, setting, and data source

This retrospective cross-sectional study was conducted in Jeju Special Self-governing Province, the largest island in South Korea (1,848 km²), with a population of approximately 580,000. The Jeju EMS is a fire-based single-tiered intermediate service-level system with a single dispatch center, four fire stations, 29 ambulances, and 168 EMS providers. In 2012, 36,995 calls were received and 30,762 patients were transported to hospitals.

The ambulance run-sheet data are from a nationwide, prospective registry of all EMS in South Korea since 2006. These data include basic ambulance operation information. EMS providers recorded the EMS run-sheets and the cardiac arrest registry for all EMS-assessed OHCA cases. A detailed description of the Korean EMS system and the quality control of the National OHCA Registry has been provided in previous studies [18-21].

We reviewed all ambulance run-sheet data from January 2010 to September 2018 and merged them with geocoded incident data, which included coordinates (i.e., latitude and longitude) for the locations of the incident.

An EMS-assessed OHCA was defined as a case in which the first symptom was recorded as cardiac arrest or respiratory arrest, or in which cardiopulmonary resuscitation (CPR) was performed, according to the ambulance run sheet data. We calculated the OHCA ratio (EMS-assessed OHCA cases/total EMS incidents) for each grid and used these OHCA ratio data to show the resulting regular distribution model, which allowed us to identify hot- and cold-spots. Hot spot were defined as grids with a z-score above 1.96 and 2.58, and cold spot were defined as grids with a z-score below -1.96 and -2.58. We expressed the hot- and cold-spot grids of the z-score Getis-Ord Gi* OHCA ratio on a map.

3. Statistical analysis

Moran’s I was calculated to determine the overall spatial interdependence among regions, and a spatial weighting matrix was used to define a contiguity matrix. Getis-Ord Gi* was calculated to capture locally heterogeneous spots with high spatial autocorrelation, even when global spatial autocorrelation was present.

4. Ethics statements

The study complied with the Declaration of Helsinki, and its protocol was approved by the Institutional Review Board (IRB) of Jeju National University Hospital, which waived the need for informed consent (IRB No. JEJUNUH 2020-04-013). Written informed consent was not required because patient data were not included in the manuscript.

RESULTS

1. Preparing spatial data for the predefined area

We used administrative district shapefiles to analyze underserved areas for EMS-assessed OHCA in Jeju, South Korea. The data included administrative boundaries, and only the districts within Jeju were extracted. We created a hexa-grid of 1 km in the defined zone range, and only grids that overlapped with Jeju were selected. Finally, 1,903 overlapping grids were analyzed. The process of preparing the spatial data is shown in Fig. 1.

2. Geocoding of EMS incidents using text-based addresses

During the study period, 366,155 EMS incidents occurred, of which 6,806 EMS-assessed OHCA were identified. The number of EMS-assessed OHCA cases and EMS incidents is summarized in Table 1. The match rates from the geocoding process of text-based incident addresses are presented in Table 2, and the geocoded addresses were incorporated into the geospatial mapping shown in Fig. 2.

3. Overlay and hot/cold spot analysis in each grid

We extracted the number of EMS incidents and EMS-assessed OHCA for each grid using a grid-based polygon layer and the point of the incident layer. We also calculated OHCA ratios (EMS-assessed OHCA cases/total EMS incidents) for each grid. Descriptive statistics for EMS incidents, EMS-assessed OHCA, and OHCA ratios in each grid are shown in Table 3.

Because the OHCA ratio in each grid was suitable for both spatial and non-spatial analyses, we performed a Moran’s I test to determine whether the residuals of the non-spatial analysis model were spatially correlated. We then defined a contiguity-based spatial weighting matrix, meaning that the share a border Moran’s I statistic (χ²=95.03, P<0.001) was significant, indicating that the residuals from the non-spatial model were correlated with those of nearby residuals.

We computed Getis-Ord Gi* statistics using a binary spatial weight matrix with a threshold distance of 10 km after confirming that the spatial pattern was not randomly distributed. The mean distance between two grids with OHCA was approximately 28.0±15.8 km, with the minimum and maximum distances of 1 km and 97.066 km, respectively. As shown at Table 4, the number of hot spot grids with median OHCA ratios at the 1.96 and 2.58 levels was 106 and 12, respectively. On the other hand, the number of cold spot grids at -1.96 and -2.58 levels was 19 and 0, respectively. Figs. 3 and 4 show a map of the z-score of the Getis-Ord Gi* for the OHCA ratio.

DISCUSSION

In this study, we introduced a geospatial methodology to identify underserved areas for EMS-assessed OHCA by combining geocoding of EMS-assessed OHCA locations, statistical information, and hexa-grid-based hot- and coldspot geospatial analysis (Getis-Ord Gi* statistic). We identified 118 and 19 grids as EMS-assessed OHCA hot- and cold-spots, respectively. Compared to the 473 grids in the first quantile of the OHCA ratio, the 118 OHCA hot spot grids provided more reliable and practical information for EMS providers. Furthermore, these findings highlight the need for further studies into why these areas are at high risk for EMS-assessed OHCA.

Previous research has shown a positive association between survival outcomes and rapid response times to BLS and ACLS in OHCA patients [22-25]. However, most studies in the field of OHCA have presented only statistical data without incorporating geospatial information. As such, they could not provide enough information to EMS providers to identify underserved areas for actual OHCA. Several studies using GIS have been conducted on patients with OHCA. In Singapore, Bayesian conditional autoregressive spatial models were used to examine predictors at the development guide plan level and identify high-risk areas for OHCA between 2001 and 2011. They identified ten high-risk residential areas with a low rate of bystander CPR and a high risk of OHCA [14]. Another study in Singapore examined the geographic epidemiology of prehospital cardiac arrest using GIS technology and assessed the propriety of deploying a public-access defibrillation (PAD) program [11]. A study in New York used ArcGIS and kernel density analysis to identify areas with the highest density of OHCA cases and the lowest rate of bystander CPR [16]. In Taiwan, a study reviewed OHCA registry data and geospatial variables using GISs to analyze spatial variations and calculate EMS transport distance from the scene to the hospital to identify risk factors associated with OHCA survival [26]. In Stockholm, a study proposed developing an analytical tool to evaluate changes in EMS system design and their effects on OHCA outcomes using geographic information. Prior to investing in EMS design changes, OHCA survival rates in the Stockholm area were estimated by evaluating the EMS system design changes and predicting OHCA outcomes. This study led to an optimized EMS system design [9].

Geospatial information is closely related to EMS systems and influences the location and use of PADs, ambulances, fire stations, and emergency rooms. This affects the outcomes of OHCA patients [10,12-15]. The studies conducted in Singapore and New York used GIS information to identify areas of incidence and poor OHCA outcomes. Another study in Taiwan used the Kriging method to calculate transport distances and demonstrated a relationship between these distances and OHCA outcomes. In contrast, we used a GIS system to calculate the OHCA/EMS incidence ratio, which distinguishes our study from previous research. The strengths of this proposed approach are as follows. First, we calculated the incidence (OHCA ratio) by joining attributes within a specific area, not the number of OHCA occurrences. Second, sophisticated spatial analysis tests, such as Moran’s I and Getis-Ord Gi* statistics, were performed instead of merely illustrating the OHCA ratio distribution to identify hot- and cold-spots. The third strength of this study is that it used a single EMS run-sheet dataset from one island with few external factors from January 2010 to September 2018.

The first limitation of this study is that we analyzed only incidents related to underserved areas for EMS-assessed OHCA without considering other factors. However, the main purpose of this study was to demonstrate differences by applying spatial analysis methods to EMS systems; therefore, we focused primarily on variables related to geospatial information and EMS utilization. Second, we did not include demographic factors such as population size, age, and the proportion of elderly individuals. These variables are likely associated with EMS and OHCA incidence rates. In future studies, we could demonstrate the association between sociodemographic factors and underserved areas for EMS, using spatial analysis. The final limitation is the lack of information on OHCA outcomes. Through further study, we need to investigate the relationship between geographical conditions and OHCA outcomes in EMS systems.

As the next step in this study, we plan to obtain spatial data containing various geographical and socioeconomic attributes, add the data to the spatial analysis layer, and analyze the spatial regression of these factors associated with underserved areas. To achieve this, it is essential to produce various types of spatial information and integrate GIS information into the EMS. This study could provide a concrete basis for assessing underserved areas for EMS in Jeju, which, in turn, could be utilized to make policy decisions, such as prioritizing emergency medical resource allocation for optimal prehospital care.

CONCLUSION

We aimed to present a geospatial methodology for defining and extracting underserved areas for EMS-assessed OHCA using a hexa-grid-based hot- and cold-spot analysis. A total of 118 hot spot grids were identified with a high incidence of EMS-assessed OHCA on Jeju Island. This GIS-based health needs assessment could help communities and decision-makers prioritize EMS resource allocation for optimal prehospital care.

Notes

ACKNOWLEDGEMENTS

None.

CONFLICT OF INTEREST

The author reports no conflict of interest.

FUNDING

None.

Figure 1.

Preparation of spatial data of the predefined area. (A) Jeju administrative district layer. (B) Creation of hexa grids (1×1 km) in the defined areas. (C) Selection of only grids (yellow) overlapped with Jeju administrative district layer. (D) Extraction of 1,903 overlapped grids layer.

Figure 2.

Geospatial mapping of geocoded addresses. (A) EMS incidents (black dots). (B) EMS-assessed OHCAs (red dots). EMS: emergency medical service, OHCA: out of hospital cardiac arrest.

Figure 3.

Z-value of EMS-assessed OHCA ratio in each grid. EMS: emergency medical service, OHCA: out of hospital cardiac arrest.

Figure 4.

Hot and cold spots of EMS-assessed OHCA in Jeju (satellite image). EMS: emergency medical service, OHCA: out of hospital cardiac arrest.

Table 1.

The number of EMS-assessed OHCA and EMS incidents per year

Year	2010	2011	2012	2013	2014	2015	2016	2017	2018^*	Total
EMS-assessed OHCA	523 (1.4)	538 (1.8)	585 (1.9)	793 (2.0)	841 (2.0)	850 (1.8)	902 (1.8)	972 (1.9)	802 (2.0)	6,806 (1.9)
EMS incidents	37,334	29,916	30,762	39,056	42,366	46,242	50,075	50,271	40,133	336,155

Values are presented as number (%) or number.

EMS: emergency medical service, OHCA: out of hospital cardiac arrest.

^* Only 9 months from January to September included in 2018.

Table 2.

The match rates from the geocoding process

Match	Value
Exact address match	263,659 (72.0)
Closet address match	70,488 (19.3)
Street/city centroid match	32,008 (8.7)
Total	366,155 (100.0)

Values are presented as number (%).

Table 3.

Descriptive statistics of EMS incidents, EMS-assessed OHCA, and OHCA ratio in each grid

Total grid	EMS incidents	EMS-assessed OHCA	OHCA ratio^*
Mean±standard deviation	192.3±671.6	2.6±12.0	0.021±0.059
Median (p25, p75)	46 (11, 138)	1 (0, 3)	0.009 (0.000, 0.025)
Minimum	1	0	0
Maximum	9,507	173	1

EMS: emergency medical service, OHCA: out of hospital cardiac arrest.

^* EMS-assessed OHCA cases/total EMS incidents.

Table 4.

Getis-Ord Gi* statistics of the OHCA ratio in each grid

z-score	Value
≥2.58	12 (0.6)
1.96≤z<2.58	106 (5.6)
-1.96<z<1.96	1,766 (92.8)
-2.58<z≤-1.96	19 (1.0)
≤-2.58	0 (0.0)
Total	1,903 (100.0)

Values are presented as number (%).

OHCA: out of hospital cardiac arrest.

REFERENCES

1. Eisenberg MS, Bergner L, Hallstrom A. Cardiac resuscitation in the community. Importance of rapid provision and implications for program planning. JAMA 1979;241:1905-7.

2. Gold LS, Fahrenbruch CE, Rea TD, Eisenberg MS. The relationship between time to arrival of emergency medical services (EMS) and survival from out-of-hospital ventricular fibrillation cardiac arrest. Resuscitation 2010;81:622-5.

3. Mullie A, Van Hoeyweghen R, Quets A; The Cerebral Resuscitation Study Group. Influence of pre-CPR conditions on EMS response times in circulatory arrest. Resuscitation 1989 17 Suppl:S45-51. discussion S199.

4. Eisenberg MS, Bergner L, Hallstrom A. Out-of-hospital cardiac arrest: improved survival with paramedic services. Lancet 1980;1:812-5.

5. Eisenberg MS, Copass MK, Hallstrom AP, Blake B, Bergner L, Short FA, et al. Treatment of out-of-hospital cardiac arrests with rapid defibrillation by emergency medical technicians. N Engl J Med 1980;302:1379-83.

6. Valenzuela TD, Roe DJ, Cretin S, Spaite DW, Larsen MP. Estimating effectiveness of cardiac arrest interventions: a logistic regression survival model. Circulation 1997;96:3308-13.

7. Valenzuela TD, Roe DJ, Nichol G, Clark LL, Spaite DW, Hardman RG. Outcomes of rapid defibrillation by security officers after cardiac arrest in casinos. N Engl J Med 2000;343:1206-9.

8. White RD, Asplin BR, Bugliosi TF, Hankins DG. High discharge survival rate after out-of-hospital ventricular fibrillation with rapid defibrillation by police and paramedics. Ann Emerg Med 1996;28:480-5.

9. Sund B. Developing an analytical tool for evaluating EMS system design changes and their impact on cardiac arrest outcomes: combining geographic information systems with register data on survival rates. Scand J Trauma Resusc Emerg Med 2013;21:8.

10. Ates S, Aydınoglu AC, Coskun MZ. Determining optimum ambulance locations for heart attack cases with GIS. In: ESRI International User Conference; 2011 Jul 11-15; San Diego (CA), USA.

11. Ong ME, Tan EH, Yan X, Anushia P, Lim SH, Leong BS, et al. An observational study describing the geographic-time distribution of cardiac arrests in Singapore: what is the utility of geographic information systems for planning public access defibrillation? (PADS Phase I). Resuscitation 2008;76:388-96.

12. Huang CY, Wen TH. Optimal installation locations for automated external defibrillators in Taipei 7-Eleven stores: using GIS and a genetic algorithm with a new stirring operator. Comput Math Methods Med 2014;2014:241435.

13. Peleg K, Pliskin JS. A geographic information system simulation model of EMS: reducing ambulance response time. Am J Emerg Med 2004;22:164-70.

14. Ong ME, Wah W, Hsu LY, Ng YY, Leong BS, Goh ES, et al. Geographic factors are associated with increased risk for out-of hospital cardiac arrests and provision of bystander cardio-pulmonary resuscitation in Singapore. Resuscitation 2014;85:1153-60.

15. Warden CR, Daya M, LeGrady LA. Using geographic information systems to evaluate cardiac arrest survival. Prehosp Emerg Care 2007;11:19-24.

16. Lerner EB, Fairbanks RJ, Shah MN. Identification of out-of-hospital cardiac arrest clusters using a geographic information system. Acad Emerg Med 2005;12:81-4.

17. Lerner EB, Billittier AJ 4th, Sikora J, Moscati RM. Use of a geographic information system to determine appropriate means of trauma patient transport. Acad Emerg Med 1999;6:1127-33.

18. Park YM, Shin SD, Lee YJ, Song KJ, Ro YS, Ahn KO. Cardiopulmonary resuscitation by trained responders versus lay persons and outcomes of out-of-hospital cardiac arrest: a community observational study. Resuscitation 2017;118:55-62.

19. Ro YS, Shin SD, Lee YJ, Lee SC, Song KJ, Ryoo HW, et al. Effect of dispatcher-assisted cardiopulmonary resuscitation program and location of out-of-hospital cardiac arrest on survival and neurologic outcome. Ann Emerg Med 2017 69:52-61. e1.

20. Ro YS, Shin SD, Song KJ, Lee EJ, Kim JY, Ahn KO, et al. A trend in epidemiology and outcomes of out-of-hospital cardiac arrest by urbanization level: a nationwide observational study from 2006 to 2010 in South Korea. Resuscitation 2013;84:547-57.

21. Shin SD, Kitamura T, Hwang SS, Kajino K, Song KJ, Ro YS, et al. Association between resuscitation time interval at the scene and neurological outcome after out-of-hospital cardiac arrest in two Asian cities. Resuscitation 2014;85:203-10.

22. Bruni-Fitzgerald KR. Review of article: early cardiopulmonary resuscitation in out-of-hospital cardiac arrest. Hasselqvist-Ax I, Riva G, Herlitz J, Rosenqvist M, et al (N Engl J Med 2015;374:2307-2315). J Vasc Nurs 2015;33:131.

23. Hasselqvist-Ax I, Riva G, Herlitz J, Rosenqvist M, Hollenberg J, Nordberg P, et al. Early cardiopulmonary resuscitation in out-of-hospital cardiac arrest. N Engl J Med 2015;372:2307-15.

24. Larsen MP, Eisenberg MS, Cummins RO, Hallstrom AP. Predicting survival from out-of-hospital cardiac arrest: a graphic model. Ann Emerg Med 1993;22:1652-8.

25. Nagao K. Chest compression-only cardiocerebral resuscitation. Curr Opin Crit Care 2009;15:189-97.

26. Chen CC, Chen CW, Ho CK, Liu IC, Lin BC, Chan TC. Spatial variation and resuscitation process affecting survival after out-of-hospital cardiac arrests (OHCA). PLoS One 2015;10:e0144882.