Spatial Accessibility to Healthcare Facilities: GIS-Based Public–Private Comparative Analysis Using Floating Catchment Methods

Abstract

Healthcare accessibility is among the most critical challenges affecting millions, reflecting profound geospatial disparities in Latin America. This study aims to evaluate healthcare service geospatial accessibility patterns, comparing the geospatial coverage between public and private healthcare facilities in Santiago district, Panama. We first apply the Two-Step Floating Catchment Area (2SFCA) method and its extended variant (E2SFCA) to calculate geospatial accessibility indexes at public and private healthcare facilities. We then use Getis–Ord Gi* and Local Moran geospatial statistical analysis to identify significant clusters of high and low accessibility. The results reveal that public healthcare facilities still offer higher geospatial coverage than private healthcare facilities, with higher geospatial accessibility in the central zone and lower geospatial accessibility in the south zone of Santiago. These findings highlighted the location of new healthcare facilities in zones with lower geospatial accessibility coverage. This study provides reproducible methodological tools for other geographical contexts. It also contributes to improving decision-making and formulating public policies to reduce spatial disparities in healthcare services in Panama and other Caribbean and Latin American countries.

1. Introduction

Access to healthcare facilities is a fundamental right for all citizens [1]. The World Health Organization (WHO) defines healthcare facilities as essential for delivering high-quality, safe, and climate-resilient care, crucial for achieving universal health coverage. These facilities, such as hospitals and clinics, are considered the core of any health system [2]. From the urban infrastructure perspective, healthcare facilities contribute to the physical and social well-being of the population [3,4]. The spatial distribution of these services influences not only health outcomes but also territorial cohesion, urban equity, and residents’ quality of life [5,6]. As cities expand, ensuring that all residents have equitable access to healthcare services becomes a significant concern for policymakers and urban planners. Therefore, measuring and evaluating geographic accessibility to healthcare facilities through GIS-based approaches becomes a relevant tool for identifying underserved areas and guiding public investment decisions [7,8,9].

Accessibility is defined as the ability of residents to reach medical facilities within a reasonable travel time, enabling quality care and disease prevention [10,11]. Some authors suggest that accessibility encompasses both the availability and acceptability of medical facilities [1,12]. According to Luo and Wang (2003) [13], accessibility refers to the ease with which people can reach healthcare service locations from a given point. Moreover, the spatial distribution of healthcare professionals and the residences of patients affects access to healthcare services due to uneven spatial distribution [13,14]. For instance, Le et al. (2022) [15] evaluated healthcare accessibility, using distance-based models that consider elements such as street networks and the distribution of healthcare facilities.

1.1. Literature Review

Previous studies have examined spatial accessibility to healthcare services in detail through geospatial tools. These investigations have encompassed a range of healthcare facilities, from hospitals and clinics [1,15,16], oncology care centers [17,18], physiotherapy services [10], ambulatory care [7], and emergency healthcare facilities [19,20]. Additionally, other authors have analyzed diagnostic and COVID-19 vaccination sites [18,21,22]. A recurring conclusion of these studies is the persistence of spatial inequalities in healthcare access, with greater emphasis on urban and peri-urban areas. For instance, Xiong et al. (2022) [1] explored the accessibility, availability, and affordability of primary healthcare in Hong Kong through spatial analysis. The educational level and economic income were social determinants influencing access to primary healthcare. The study demonstrated that, although the accessibility and availability of primary care services are adequate, affordability-related difficulties persist for specific population groups. Wang and Onega (2015) [18] proposed a research framework for analyzing the accessibility to oncological care, focusing on measuring geographic and demographic disparities in access to cancer centers. Their methodology employed advanced models based on Geographic Information Systems (GIS) to quantify accessibility, considering both spatial and socioeconomic factors, and multilevel models to examine the relationship between access, service utilization, and cancer-specific mortality outcomes.

The literature identifies multiple methodological approaches for measuring spatial accessibility, primarily within the context of GIS. Systematic reviews, such as those conducted by various authors [14,16], highlight the role of these tools in disease mapping, territorial planning, and the distribution of healthcare services. One of the most widely used methods is the Two-Step Floating Catchment Area (2SFCA), which integrates supply and demand factors within a determined analysis radius [1,8,11,16,23]. Variants such as the Enhanced Two-Step Floating Catchment Area (E2SFCA) model incorporate distance decay functions for a more realistic representation [9,22,24], while the Three-Step Floating Catchment Area (3SFCA) method adds additional criteria to reflect the competition between providers or the population distribution [10,25,26]. Although the 3SFCA method has been recognized for its ability to incorporate provider competition and improved demand weighting [26,27], its application requires data on patient behavior, healthcare provider capacity, and competition parameters, which are often unavailable in developing country contexts. In this study, we focused on the 2SFCA and E2SFCA methods, which have been widely validated and remain suitable for evaluating spatial accessibility in data-limited environments, such as Panama. Future research could consider applying 3SFCA once such detailed inputs become available.

While extensive research has been conducted in high-income urban areas in developed countries, there is still a lack of investigation in most developing nations’ settings, such as Latin America. For example, some of these studies have been conducted in large cities, such as Hong Kong [1], New Zealand [12], Chicago [13], and Florida [21], where the data availability is relatively high. In this regard, some research focuses on analyzing healthcare accessibility in rural areas using the Floating Catchment Area method [10,27]. Mahuve et al. (2022) [28] applied an enhanced version of the E2SFCA method targeted to a rural area of Tanzania. Wang (2011) [29] used the 2SFCA method to determine healthcare accessibility among various ethnic groups in Toronto. These analyses have provided methodological insights. But a significant gap remains in understanding accessibility patterns in Latin American countries, particularly in rural areas with limited healthcare and population data availability. To address this, the present study applies geospatial analysis methods to assess healthcare accessibility in the Santiago de Veraguas district. Santiago de Veraguas is a mid-sized city in Panama, with a population of 109,605 inhabitants distributed across 16 districts. Therefore, this study contributes new insights from a Latin American perspective, which is a region that remains underrepresented mainly in the spatial health research literature.

1.2. Research Questions and Objective

This study proposes using the 2SFCA and E2SFCA models, which are among the most widely employed methods for measuring population accessibility to healthcare facilities. Unlike previous studies focused on large urban areas, our research offers a comparative assessment of the accessibility to public and private healthcare facilities, providing support for territorial planning in resource-limited regions. The primary objective of this study is to evaluate the geographic accessibility patterns of the population in the Santiago de Veraguas district, Panama, to both public and private healthcare facilities. In this regard, this paper addresses the following main research questions to guide our work:

• What are the spatial patterns of accessibility to public and private healthcare facilities in the Santiago de Veraguas district?
• How do geospatial accessibility differences in service coverage between public and private sectors influence the identification of underserved areas?

In this study, we implement a network analysis approach with ArcGIS Pro 3.4.3 software tools to calculate accessibility to healthcare facilities based on different travel time intervals. The results reveal better accessibility to public healthcare facilities and a higher concentration of services in the central area of the studied region.

Spatial accessibility studies of healthcare facilities in developing countries are crucial, as many of these nations lack territorial analyses of the distribution and quality of healthcare facilities. Governmental and private entities can use this research as a reference to support decision-making processes when planning the location of new healthcare facilities.

The paper is structured in the following way. First, Section 2 presents the research method undertaken in this study. Section 3 presents our results on the evaluation of the geographic accessibility patterns of the population in the Santiago de Veraguas district, Panama. In Section 4, we discuss our findings. Finally, Section 5 closes the paper by presenting our main conclusions.

2. Materials and Methods

2.1. Study Area

We selected the city of Santiago district as the study area due to its strategic role as the capital of the province of Veraguas, Panama (Figure 1). According to the latest census by the Institute of Statistics and Census (2020) [30], the city has a population of 109,605 inhabitants distributed across 16 districts, with a population density of 113 inhabitants per square kilometer. The service sector primarily constitutes the base of the economy. Major establishments include restaurants, shopping centers, pharmacies, and clinics that provide medical healthcare to the local population. The Pan-American Highway crosses the city from east to west, supplemented by a network of secondary roads facilitating intra-urban mobility. The Panamanian healthcare system classifies medical facilities as either public or private institutions. Most of the population accesses services at the Social Security Hospital and the Luis Chicho Fábrega Hospital. This research selected the Santiago district as the study site based on two primary considerations: its significant role as the administrative and service center for the country’s central region and the imperative to assess healthcare accessibility patterns within developing nation urban environments.

2.2. Datasets

The research was conducted through the integration of three complementary geospatial datasets. The first dataset consisted of census information on population settlements obtained from the Statistics and Census Bureau of the Republic of Panama. This data, corresponding to the year 2020, detailed the spatial distribution of 289 populated places within the Santiago district and its total population of 109,605 inhabitants, which were georeferenced to represent the potential demand for healthcare facilities.

The second dataset, comprising the geographic locations of healthcare facilities (both public and private), was downloaded from OpenStreetMap (OSM) using the Overpass Turbo query tool version 0.7.62.4 [31]. This platform enabled us to extract point features classified under tags such as “amenity = hospital”, “amenity = clinic”, and “amenity = healthcare = center”, which are commonly used in the OSM schema to identify medical institutions. The “hospital” and “healthcare center” amenities represent the public healthcare facilities. Meanwhile, the “clinic” amenities represent the private healthcare facilities. The OSM dataset included the name, category, and precise geographic coordinates of each facility. The number of physicians in each healthcare facility was also included, based on a field visit and a consultation of the database from the regional office of the Ministry of Health and Social Welfare in the Santiago de Veraguas district. A validation process was conducted to clean the data, remove duplicates, and verify that all facilities were within the boundaries of the Santiago district. The resulting georeferenced layer was used for subsequent spatial accessibility and cluster analyses (Figure 1a).

This study also incorporated a third dataset to integrate healthcare facilities’ supply and demand data, consisting of the OpenStreetMap street network (Figure 1b). In order to facilitate geospatial accessibility analysis, a connectivity network was built using its edges and nodes. This network allowed for the modeling of possible transit from residential areas to medical care centers (Figure 1b).

2.3. Methods

In this subsection, we introduce the methods used to undertake this research. In order to evaluate accessibility indexes to healthcare facilities, the 2SFCA and the enhanced E2SFCA methods were implemented. The 2SFCA method was used with fixed catchment areas and equal weighting to estimate the accessibility index. This method may not accurately reflect actual demand behavior because it assumes that all healthcare facilities are equally accessible, regardless of travel time and distance. To address this limitation, we additionally incorporated the E2SFCA method. This method uses a cost function that penalizes healthcare facilities located farther from the patient’s place of residence. By using both methods, our research captures accessibility index differences to healthcare facilities in the study area, considering both fixed catchment areas and weighted travel time estimations [19,32,33,34].

2.3.1. Data Gathering and Preparation

The development of accessibility models relied on three essential elements: (1) point locations of the healthcare facilities (offer of health services), (2) point locations of the geographical distribution of the population (demand of health services), and (3) a street network that simulates the communication channel between the offer and demand of health services. To optimize spatial analysis, we aggregated population data into hexagons, reducing the number of points used as demand sites in the accessibility index calculation. The hexagon size was determined following the methodology proposed by Barros, Moya-Gomez, and Gutierrez (2019) [35], resulting in a side length of 1000 m. Studies highlighting the efficiency of this geometry in spatial pattern analysis support its selection [17,36,37,38,39,40]. The choice of hexagons as the spatial unit was guided by methodological advantages: hexagonal cells reduce edge effects and directional bias, as each cell has six equidistant neighbors; the distance between the centroid of a hexagon and the neighboring centroids is the same in all directions; all neighboring hexagons have the same spatial relationship to the central hexagon; its smoother shape helps to better visualize gradual spatial changes [37,38,39,41], unlike square grids, which vary in distance and adjacency [42]. This geometry allows consistent spatial interaction modeling and uniform area comparison. In contrast, administrative units such as census tracks are often irregular in size, which can introduce bias in accessibility measurements. The use of a regular hexagonal grid ensured statistical consistency across the study area and is increasingly used in spatial epidemiology and urban planning studies [43,44]. In addition to the hexagon-based model, we incorporated the street network as a key element in developing accessibility indexes. We constructed a street network using street segments and nodes extracted from OpenStreetMap via the Overpass Turbo online application version 0.7.62.4. The street network dataset included road type (highway) and direction (one-way) attributes. But it had limited availability of speed limit data (maximum speed). To enrich the network dataset, we purposefully assigned speed limits to each street segment based on each street hierarchy, according to Panama’s national traffic regulations (e.g., 40 km/h for residential roads, 80 km/h for primary roads).

The network was structured as a directed graph to preserve flow restrictions and was integrated into ArcGIS Pro 3.4.3 using the Network Analyst toolbox. The travel time for each segment was estimated based on assigned speed limits and segment length; accessibility analyses were performed using time as travel cost impedances to simulate automobile accessibility.

2.3.2. Accessibility Indexes

In this study, the accessibility indexes were calculated using the 2SFCA and E2SFCA models. These models are widely applied in academia to measure population accessibility to healthcare facilities. Both models considered travel time metrics and coverage areas to evaluate geographic accessibility patterns in the Santiago of Veraguas district for public and private healthcare facilities. We established comparisons between different facility types. The accessibility models were applied to all healthcare facilities and then separately to public and private facilities. The difference between the accessibility to public and private healthcare facilities in Panama is stark. The government separates the country’s healthcare system into two distinct parts. Many people, especially those without private insurance or from low-income families, rely on public services for their healthcare needs. In contrast, private services cater to a smaller, higher-income group. When we examined the accessibility of these services, we uncovered spatial inequities in how different groups access them. We developed a comprehensive accessibility model that covers all healthcare options. This approach mirrors real-life situations where patients choose care from either sector based on urgency or location. This helps inform better urban and health planning strategies. It also answers our second research question by identifying underserved areas, regardless of their administrative classification, which reinforces the policy relevance of our findings.

The implementation used ArcGIS Pro 3.2, specifically the Network Analyst module and its service area tools. The 2SFCA method defines the service areas of healthcare facilities based on a predefined travel time threshold, considering physician availability relative to surrounding demand.

2SFCA model

The 2SFCA model is structured in two steps to obtain the accessibility index, considering the relationship between available healthcare facilities and the spatial distribution of the population. According to Kuai and Zhao (2017) [45] and Hashtarkhani (2024) [46], the procedure is as follows:

Step 1: For each healthcare center, identify the population covered within a predefined travel time threshold. The healthcare-to-demand ratio is calculated using the following equation:

$$R_j={\displaystyle \frac{S_j}{{\sum }_{k\in \left\{d_{kj}\le d_0\right\}}{P}_k }},$$

(1)

where P_k represents the population at location k within the coverage area j (d_kj ≤ d₀), S_j is the capacity of the healthcare center at location j, and d_kj is the travel time between k and j.

Step 2: For each demand location i, identify all healthcare centers j within the specified travel time threshold (d₀) and sum the healthcare-to-demand ratios (R_j) obtained in Step 1.

$$A_i^F {\sum }_{j\in \left\{d_{ij}\le d_0\right\}}{R}_{j },$$

(2)

where $A_i^F$ represents the accessibility at demand location i based on the 2SFCA method. R_j represents the healthcare-to-demand ratio for healthcare centers j within the service areas of demand site i (ie, d_ji ≤ d₀) and d_ij is the travel time threshold between i and j.

To implement the 2SFCA model, we used the locations of public and private healthcare centers and the number of physicians providing services. The potential demand was represented by centroids containing population data for settlements in the Santiago district. A 15 min automobile travel threshold was applied, considering speed limit data from the street network. Previous accessibility studies of healthcare centers have demonstrated the validity of using a 15 min threshold [1,7,15]. Furthermore, the 15 min city concept has been developed in academic research as a criterion for evaluating urban sustainability and resilience [47].

E2SFCA model

The E2SFCA model incorporates three influence zones defined by different travel time intervals (10, 15, and 20 min). This enhancement accounts for the decline in accessibility with distance using a Gaussian function that weights access according to the influence zone. The selection of these intervals is based on previous studies on healthcare service accessibility [15,33]. The E2SFCA method utilized travel time intervals of 10, 15, and 20 min, chosen based on the geographic and urban characteristics of the Santiago district, a medium-sized city featuring short travel distances. Using longer intervals, like 30 min, would have exceeded the spatial extent of the district and resulted in an overestimation of service areas.

The inclusion of a 15 min interval aligns with recommendations in the literature that suggest this threshold is a reasonable standard for urban healthcare access [48,49], especially for primary or general outpatient services. The 10, 15, and 20 min intervals provide finer granularity to assess potential inequalities in spatial accessibility, capturing both core and fringe areas of coverage. All other parameters are consistent with those used in the 2SFCA model. According to Luo and Qi (2009) [33] and Hashtarkhani (2024) [46], the E2SFCA calculation procedure is as follows:

Step 1: For each healthcare center j, generate three service areas based on 10, 15, and 20 min travel times. All population centroids (k) within each service area (D_r) are identified, and the healthcare-to-demand ratio (R_j) is calculated using the following equation:

$$\begin{gathered} R_j={\displaystyle \frac{S_j}{\sum _{k\in }\left\{d_{kj}\in D_r\right\}{P}_k{W}_r}} \\ ={\displaystyle \frac{S_j}{\sum _{k\in }\left\{d_{kj}\in D_1\right\}{P}_k{W}_1+\sum _{k\in }\left\{d_{kj}\in D_2\right\}{P}_k{W}_2+\sum _{k\in }\left\{d_{kj}\in D_3\right\}{P}_k{W}_3}} \end{gathered}$$

(3)

where P_k represents the demand at location k within the service areas of j (d_kjD_r), S_j is the number of healthcare services at location j, d_kj is the travel time between k and j. D_r is the rth catchment zone (r{1,2,3}) within the service areas. W_r represents the weight for each zone, calculated using the Gaussian function to account for distance decay in accessibility to healthcare center j.

Step 2: For each population centroid i, identify all healthcare centers j within 10, 15, and 20 min travel times and sum the weighted healthcare-to-demand ratios R_j calculated in Step 1.

$$A_i^F={\sum }_{j\in \left(d_{ij}\le D_r\right)}{R}_j{W}_r={\sum }_{j\in \left(d_{ij}\le D_1\right)}{R}_j{W}_1+{\sum }_{j\in \left(d_{ij}\le D_2\right)}{R}_j{W}_2+{\sum }_{j\in \left(d_{ij}\le D_3\right)}{R}_j{W}_3,$$

(4)

where $A_i^F$ represents accessibility at demand location i, R_j is the healthcare-to-demand ratio at location j within the service areas of i (ie, d_kj D_r), and d_ij is the travel time between i and j. The same Gaussian functions from Step 1 were applied to account for distance decay.

2.3.3. Spatial Clustering Analysis

To spatially interpret the results obtained from accessibility indexes calculated using the 2SFCA and E2SFCA methods, spatial analysis techniques were applied to identify significant clustering patterns. First, the Getis–Ord Gi* statistic was employed, which enables the detection of significant concentrations of high or low accessibility values, namely, the identification of hot spots (high accessibility clusters) and cold spots (low accessibility clusters) through the analysis of z-scores and statistical significance levels [50,51]. This tool proves helpful in highlighting critical areas where accessibility is considerably low and, therefore, requires priority attention in territorial planning.

Complementarily, the Anselin Local Moran’s I local autocorrelation statistic, also known as LISA (Local Indicators of Spatial Association), was applied to examine whether extreme values identified in the accessibility index are spatially correlated with their immediate neighbors [21,52,53]. The Moran I index allows for distinguishing between clusters of high accessibility surrounded by similar areas (high–high), low accessibility surrounded by low accessibility (low–low), or atypical spatial patterns (high–low, low–high). It also provides a more detailed interpretation of territorial structure [54]. The combination of both statistical analyses, combined with the 2SFCA and E2SFCA results, provides evidence of the unequal distribution of geographic accessibility, facilitating the identification of gaps and the targeting of more equitable public policies [54,55].

3. Results

3.1. Service Areas

The service area analysis of the public and private healthcare facilities in the Santiago de Veraguas district considered travel times of 10, 15, and 20 min. The study evaluated three scenarios, all medical facilities in the district, exclusively public, and private healthcare facilities, independently.

The spatial distribution analysis of all healthcare facilities (Figure 2a) revealed that the central districts (Santiago Cabecera, San Martín de Porres, Nuevo Santiago, Rodrigo Luque, and Canto del Llano) demonstrated optimal coverage within a 10 min travel time. The 15 min threshold analysis indicated that the eastern and western zones (districts Nuevo Santiago and Carlos Santana) required longer travel times. In contrast, districts such as Ponuga, La Raya de Santamaría, and Los Algarrobos, located in the eastern and southern zones, required a travel time of more than 20 min to be within the service areas of a healthcare center.

The analysis of public healthcare facilities (Figure 2b) showed spatial distribution patterns similar to those observed for all healthcare facilities, with a notable variation in the Urracá district. Conversely, the service areas of private healthcare facilities (Figure 2c) exhibited lower coverage, particularly in the southern district, where coverage was nonexistent.

3.2. Comparative Analysis of Spatial Accessibility Models

Initially, the 2SFCA accessibility index was calculated for all healthcare facilities, followed by separate calculations based on the type of healthcare facilities (public or private).

Table 1. Overall values of accessibility indexes in Santiago de Veraguas district.

Healthcare	2SFCA						E2SFCA
	Mean	SD	C.V.*	Median	Min	Max	Mean	SD	C.V.*	Median	Min	Max
Public	11.6	7.8	67.1	11.8	0.0	23.4	9.1	6.0	66.0	9.5	0.0	24.2
Private	5.7	6.2	109.3	0.6	0.0	12.9	2.9	2.6	89.6	3.4	0.0	6.8

SD = Standard deviation. C.V.* = coefficient of variation.

shows the overall values of accessibility indexes in the Santiago de Veraguas district obtained with both accessibility models. According to the 2SFCA model, public and private healthcare facilities show different spatial patterns. The public healthcare facilities showed index values ranging from 0.00 to 23.36 (mean = 11.58; median = 11.82; SD = 7.77), suggesting wider and more evenly distributed accessibility.

In contrast, the private facilities had a more skewed distribution (range: 0.00–12.89; mean = 5.6; median = 0.5; SD = 6.2), reflecting concentration in central urban zones. This disparity suggests an inequity in access, particularly in rural or remote areas.

The E2SFCA accessibility index showed distinct spatial patterns. For the public healthcare facilities, the index values ranged from 0.0 to 24.2, with a mean of 9.1, a median of 9.5, and a standard deviation of 6.0, indicating broad and relatively even coverage across the facilities. In contrast, the index values for the private healthcare facilities ranged from 0.0 to 6.8, with a mean of 2.9, a median of 3.4, and a standard deviation of 2.6, indicating a more uneven distribution concentrated in central urban areas. These results suggest that public services are more evenly distributed across the study area, while private healthcare services are clustered in high-accessibility zones, with limited coverage in peripheral regions. When comparing the statistics of both models, public healthcare facilities have higher accessibility values (mean and median) than private centers. The coefficients of variation indicate moderate dispersion for public healthcare facilities (67% concerning the mean). In contrast, the accessibility to private centers shows that the relative variability is very high, reaching over 100% of the mean. In summary, the coefficient of variation shows that the accessibility to public healthcare facilities is more homogeneous than that recorded for private centers, regardless of the method used.

The analysis of all of the healthcare facilities (Figure 3a) revealed greater coverage and service availability in the district’s central zone, encompassing the districts of Santiago, Canto del Llano, San Martín de Porres, and Rodrigo Luque; this zone exhibited the highest accessibility indices. These districts have a higher population density and, therefore, a greater demand for healthcare services. Additionally, this area hosts most healthcare centers and private clinics. These districts also host most of the public services and businesses. In contrast, the northeastern, eastern, and southern districts exhibited limited accessibility indices, particularly in the districts of La Peña, Santa María, and Santiago Sur, where the healthcare facilities’ locations were more than 15 min away by car. The poor road infrastructure in these districts could be one of the reasons for this low accessibility to healthcare facilities. These zones mainly have a rural population. As a result, citizens living there struggle to access healthcare services, particularly private healthcare facilities.

The independent analysis of the public and private centers (Figure 3b,c) revealed higher accessibility in the central zone. However, the values were significantly lower compared to the indices obtained for all of the healthcare facilities (Figure 3a). The accessibility analysis for the private healthcare facilities demonstrated that the indices were lower than those obtained for the public healthcare facilities, maintaining higher accessibility patterns in the centrally located districts. Additionally, the analysis revealed low accessibility to private healthcare facilities in the southern district.

Regarding the E2SFCA index, the analysis reflected an improvement in accessibility across most of the Santiago district when considering all of the healthcare facilities, with indices ranging from 16 to 25 physicians per 10,000 inhabitants. This improvement was particularly notable in the districts of the central zone (Santiago Cabecera, San Martín de Porres, Los Algarrobos, Canto del Llano, and Urracá). Additionally, the analysis identified better accessibility in the southern district, including the districts of Ponuga and Santiago Sur (Figure 4a).

On the other hand, the accessibility index was lower when the E2SFCA algorithm was applied exclusively to the public healthcare facilities. The analysis revealed a similar distribution pattern for all healthcare facilities (Figure 4b). In the case of private healthcare facilities, the accessibility indices were low, with values ranging from 5 to 10 physicians per 10,000 inhabitants. The accessibility in the southern districts remained low (Figure 4c).

Figure 5 illustrates the distribution of the 2SFCA and E2SFCA accessibility indices calculated for public and private healthcare facilities. The central districts exhibited the highest values in both models, while the southern districts had better access to public healthcare facilities. The E2SFCA model reflected higher coverage indices in the southern zone for public healthcare facilities without affecting coverage in the central zone (Figure 5b).

3.3. Spatial Pattern Analysis

Figure 6 presents the cluster formation identified through Getis–Ord Gi* and Moran’s I analysis. This analysis compared the accessibility indexes calculated using the 2SFCA-E2SFCA models for all healthcare facilities. In the spatial distribution of the 2SFCA model, the district’s central zone exhibited high-accessibility clusters (hot spots). In contrast, the southern districts (Ponuga, Santiago Sur) displayed clusters with low accessibility levels (Figure 6a).

The E2SFCA model revealed distinct spatial patterns, with a reduction in the size of the central cluster and the emergence of two new clusters in the southern districts (Figure 6b). The comparison between the 2SFCA and E2SFCA models (Figure 6c) demonstrated that the central districts maintained their hot spot status. At the same time, Ponuga and part of Santiago Sur transitioned from cold spots to hot spots.

The Local Moran’s I analysis (Figure 6d,e) represented the distribution of clusters and outliers detected in the Santiago district. The accessibility values obtained from the 2SFCA model showed a high-accessibility cluster in the central districts (Santiago Cabecera, San Martín de Porres, Canto del Llano, and Rodrigo Luque). Conversely, the indices obtained with the E2SFCA model indicated the presence of two clusters: one in the central zone and another in the district of Ponuga (Figure 6e). Figure 6f illustrates the transitions in cluster types recorded with both models, maintaining the central zone’s clustering while the southern area transitioned from low to high values.

4. Discussion

This study evaluated the geographic accessibility patterns of the Santiago of Veraguas district population to public and private healthcare facilities by applying spatial accessibility models to all healthcare facilities and then differentiating between public and private facilities. Impedance factors informed this analysis, including the travel time through the road network and permitted speed limits. This study employed the 2SFCA and E2SFCA spatial accessibility indexes to compare the availability of accessible healthcare facilities for the population. Using both methods enabled a more comprehensive evaluation of healthcare accessibility in Santiago de Veraguas by incorporating different assumptions. The E2SFCA method, with its distance decay functions, offers a more refined perspective that may reveal patterns missed by the traditional 2SFCA. This comparison underscores how assumptions shape the spatial distribution of accessibility.

The findings reveal that the central districts of the study region exhibited the highest accessibility indexes, which aligns with previous research indicating more favorable access in urban areas compared to peripheral zones [4,25,46]. The central and northern districts—where accessibility is higher—coincide with the historic urban core and the city’s primary economic corridors, where institutional healthcare supply is concentrated.

When analyzing the districts with the lowest coverage, it was found that their access was limited to public healthcare facilities, as the distance to private healthcare facilities exceeded 20 min of travel time. The southern sector of the Santiago de Veraguas district, which exhibits the lowest accessibility indexes in both the public and private healthcare systems, is characterized as a rural area with a low population density and a lack of private healthcare facilities. The unequal distribution of these facilities generates areas with limited accessibility in the south. The results obtained for the southern region of the district suggest the necessity to create new public or private healthcare facilities or the expansion of existing facilities to ensure equitable access to healthcare services.

When comparing the accessibility indexes between the public and private healthcare facilities, the results suggest that the public healthcare facilities provide better coverage. The dispersal of health posts—at least one in each district capital—as opposed to the concentration of private clinics in the district centers is probably the cause of this phenomenon. In addition, the public healthcare system includes two hospitals with greater physician availability in the Santiago de Veraguas districts. This significantly enhances their capacity to deliver healthcare facilities and increases their influence in the accessibility model.

Townships located in the district’s center have easier access to private healthcare facilities than those in the south. This trend emphasizes how inhabitants in outlying areas must drive farther to the center area, home to most private healthcare. Peripheral communities’ limited access to healthcare services may result in subpar treatment, which could lower the standard of living for the local people. Rural areas should be given priority when building and allocating healthcare facilities because they have less infrastructure and fewer doctors to meet demand.

Various researchers have conducted accessibility studies in other Latin American cities, where territorial inequalities strongly influence healthcare accessibility. Cortes (2021) [55] found an unequal distribution of accessibility to healthcare services in Chile associated with socioeconomic factors. Cuervo et al. (2022) [56] in Cali, Colombia, identified significant accessibility inequalities related to socioeconomic segregation and the concentration of services in high-income areas. In their work, Bascuñán and Quezada (2016) [57] suggest that regions with lower access to medical facilities tend to have populations with lower incomes. These studies support the interpretation that, in Santiago, the extended reach of public services may reflect state-driven equity strategies. In contrast, private services follow market logic, clustering in higher-income urban sectors. Although socioeconomic variables were not considered in our research, the inclusion of such demographic indicators in future analyses would allow for a better understanding of how structural factors influence healthcare accessibility.

This research might be relevant for the Republic of Panama as well as the scientific community from other developing countries in Latin America, the Caribbean region, and around the world. To the best of our knowledge, this work represents the most comprehensive spatial accessibility study of healthcare facilities in the Republic of Panama and any other Central American country and the Caribbean region ever undertaken. Therefore, the results of the spatial accessibility study to healthcare facilities offer a roadmap to support decision-makers in planning future allocations of healthcare facilities, which contributes to reducing the gap to health services in underserved areas of developing countries. The results of this study may also be helpful for international agencies and private investors that are supporting developing countries such as the Republic of Panama to enhance their access to healthcare services and to implement future strategies to achieve Sustainable Development Goal (SDG) 3, also known as “Good Health and Well-being”.

There are some limitations to our research approach. First, socioeconomic and demographic data availability in developing countries remains a challenge, particularly regarding healthcare statistics for private facilities. Concerning population data, the most detailed level of data disaggregation available corresponds to districts, preventing the analyses at the neighborhood, block, or parcel level. Using census data gathered into hexagonal units helped to reduce this restriction and enable more accurate calculations of the accessibility index.

Another major drawback is the omission of important factors, including the availability of private vehicles, public transit, medical specialties, and the percentage of insured and uninsured populations. The results, limited by the absence of these variables, should be interpreted within the methodological framework of this study. Moreover, it is important to highlight that the dataset used in this study has a low temporal resolution. In particular, the population, infrastructure, and road network datasets only represented static snapshots of the current settings in the Santiago de Veraguas district. This restricts the study’s ability to assess temporal variations in accessibility to healthcare facilities—such as changing conditions during rush hours or seasonal fluctuations.

Future studies should consider integrating time-sensitive data to improve the dynamic analysis of spatial access to healthcare services. For example, historical or real-time traffic data could be incorporated. Furthermore, adding data on floating populations from social media or cell phone records may increase the precision of spatial analysis. In addition, exploring the application of machine learning algorithms that include socioeconomic and infrastructure variables in accessibility modeling is suggested.

5. Conclusions

In this paper, we report on the implementation of the 2SFCA and E2SFCA models to obtain the patterns of spatial accessibility to public and private healthcare facilities in the Santiago de Veraguas district. For practical reasons, the 2SFCA and E2SFCA models were implemented to calculate accessibility indexes for each type of healthcare facility, using the travel time through the street network and catchment areas between the supply and demand locations of healthcare facility services, and, thus, answered the following research question: what are the patterns of spatial accessibility to public and private healthcare facilities in the Santiago de Veraguas district?

Our overall results revealed that public healthcare facilities (23.4) present a higher level of spatial accessibility than private healthcare facilities (12.9). Our findings also showed that public healthcare facilities offer broader and more balanced spatial coverage across both urban and peripheral zones. In contrast, private healthcare facilities are mainly concentrated in central urban areas, resulting in more fragmented accessibility patterns.

This paper also aimed to identify how geospatial accessibility differences in service coverage between public and private sectors influence the identification of underserved areas. Our study also reveals that some areas with lower private healthcare service coverage are located in the peripheral zone of the study area. The results also show that public healthcare facilities cover a greater portion of the territory; however, underserved areas are still found. Planners can use these differences as a reference to address areas with low accessibility to healthcare facilities. Furthermore, the spatial distribution of the accessibility indices obtained from the 2SFCA and E2SFCA models was analyzed using Getis–Ord Gi* and Local Moran statistics to identify high- and low-accessibility spatial clusters. According to this study, the central portion of the district has the best coverage; the southern zones show the lowest accessibility indexes.

The results of this research are particularly relevant for legislators, urban designers, and government agencies, as they contribute to increasing fairness in access to healthcare. New healthcare facilities should be established in areas with low accessibility or the number of physicians should be increased in existing facilities.

This study also opens new avenues for research. One approach is to incorporate historical traffic data into the network using big data sources, such as TomTom or the Google Maps API. Another line of research could be the use of GTFS files, which store information on travel times and transit schedules. This approach would enable the assessment of accessibility to services by public transport and allow dynamic temporal analyses throughout the day, facilitating the evaluation of accessibility variations according to traffic patterns and public transport availability.

Finally, this methodology can be applied to cities in developing countries to evaluate and compare access to healthcare services, incorporating socioeconomic variables into the analyses.