*Article* **Assessing Health Resources Equipped with Hemodynamic Rooms in the Portuguese-Spanish Borderland: Cross-Border Cooperation Strategies as a Possible Solution**

**José Manuel Naranjo Gómez 1,2, Rui Alexandre Castanho 2,3,4,\*, José Cabezas Fernández 2,5 and Luís Loures 2,6**


**Abstract:** Portugal and Spain share one of the greatest European borderland areas. This fact has direct impacts on a large territory and consequently on the communities' living in it. Still, even if the border areas represent an essential fraction of the territory, planning policies have not resulted in specific cooperation programs that could enable sharing general leisure and recreation assets and infrastructures and collaboration in critical domains—i.e., the case of the health sector. The present study aims to assess the territorial accessibility to the hemodynamic rooms by the potential population of the Spanish-Portuguese transition areas that may suffer an acute myocardial infarction. Contextually, this study employed a spatial interaction model based on the three-step floating catchment area method (method-3SFCA). By applying these methods, it was possible to develop a map of accessibility to health infrastructures equipped with hemodynamics rooms on both sides of the border that may answer the Spanish-Portuguese border populations' needs. Besides, while granting valuable information for decision-makers regarding the need to develop new infrastructures to guarantee that even considering cross border cooperation, everyone gets access to a hemodynamics room within the critical intervention period.

**Keywords:** cross-border cooperation; geographic information systems; Iberian borderland; strategic planning; sustainable planning

#### **1. Introduction**

Providing adequate essential services is an increasingly important issue in livelihoods, sustainability, and public policy [1]. Equality of access to these services must be achieved between different population groups regardless of social, economic, demographic, or geographical differences that in many cases lead to inequalities in the provision of these services [2–4]. Mainly it must be achieved in those services that are vital, such as the health service [5,6].

Nevertheless, in many cases, the provision of health services is not distributed equally in one region and among different population groups due to a variety of spatial and nonspatial factors [7,8]. As for non-spatial factors, these mainly significantly affect the quality of the health service offered [7,9]. However, spatial aspects can become a physical barrier that hinders adequate access to health services, depending on the separation distance where the patient needs medical assistance to the nearest hospital where he or she can be assisted [10,11]. For this reason, health service planning must make it accessible and

Loures, L. Assessing Health Resources Equipped with Hemodynamic Rooms in the Portuguese-Spanish Borderland: Cross-Border Cooperation Strategies as a Possible Solution. *ISPRS Int. J. Geo-Inf.* **2021**, *10*, 514. https://

doi.org/10.3390/ijgi10080514

**Citation:** Gómez, J.M.N.; Castanho, R.A.; Cabezas Fernández, J.;

Academic Editors: Fazlay S. Faruque and Wolfgang Kainz

Received: 3 June 2021 Accepted: 28 July 2021 Published: 30 July 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

effective for the whole of the served population [12–14]. For this reason, the location of health infrastructures has a social impact among residents of cities and towns [15,16]. In fact, a country's health care capacity affects others due to the interconnection between economic, social, technological, and political systems [17].

In this regard, considering the different political systems of Spain and Portugal, each country depends on its criterion of self-sufficiency to provide health infra-structures, producing unequal technological development. These differences are more prominent in high-tech sanitary technology equipment, mainly because it is more expensive and requires more skilled human resources. In fact, the non-existence of prior collaboration agreements or previously established protocols causes the patient to be treated primarily in the same country where he or she requires healthcare. Likewise, if the population's access to health services may in some cases be inadequate and unequal [18], this is aggravated concerning high-tech health care [19], like hemodynamic rooms that are considered representative precisely of high health care technology for three fundamental reasons. Firstly, access time to these rooms is a vital factor for the patient. Secondly, they offer urgent health benefits by treating myocardial infarction through percutaneous coronary angioplasty (PCA) which is a treatment potentially usable in the vast majority of patients with acute myocardial infarction, achieves the recanalization of the coronary artery in more than 90% of cases, with better perfusion of the infarcted area and a lower incidence of re-inclusion when compared with thrombolytic treatment [20]. Thirdly, these rooms must be integrated into a hospital [10].

The assessment and optimization of cardiovascular and hemodynamic variables is a mainstay of patient management in critically ill patients in the intensive care unit (ICU) or the operating room (OR). Therefore, it is of outstanding importance to meticulously validate technologies for hemodynamic monitoring and to study their applicability in clinical practice and, finally, their impact on treatment decisions and patient outcome [21]. Indeed, hemodynamic monitoring provides the basis for optimizing cardiovascular dynamics in intensive care medicine and anaesthesiology [22]. The hemodynamic room has the most advanced technology for the diagnosis and treatment of coronary diseases. On the one hand, the virtual histology offers differentiated and percentage information of the four components of the atheroma plaque: the fibrotic, lipid, calcium, and chorionic content. On the other hand, intracoronary ultrasound consists of a 1 mm catheter incorporating a small ultrasound emission system (transducer) at the distal end connected to a console that generates images in real-time.

The use of indicators makes it possible to assess the geographical distribution of the applicant population of these services, the transport infrastructures that allow access to them, and the provision of services offered by health infrastructures. In this sense, the proliferation of spatially disaggregated data and the increasing use of Geographic Information Systems (GIS) has led to a plethora of spatial accessibility analysis to health services, combined with criteria for quality and efficiency of the service provided [10]. In the bibliography on accessibility, a great variety of indicators that evaluate them is observed (see, for example, [23–27]). Moreover, all of them offer a measure to close the existing separation between human settlements and activities depending on the transport network made available to defining a single indicator of accessibility that includes all the approaches is nearly impossible to realize since each indicator measures this variable from a concrete point of view, to evaluate the quality of the access to the transport infrastructures and to determine strategic locations or as tools for the planning in the decision-making process [26]. However, these indicators can complement each other to provide a clearer picture about the benefits that the analyzed infrastructure will facilitate the territories affected by it [28]. Therefore, it makes it possible to analyze whether societies are inclusive [29], that all their members have access to high-tech healthcare services and impartial, and whether they have the same healthcare opportunities [30–32]. This kind of analysis has produced different approximations [18,33], including the regional availability model [34–37], kernel density models [29–31], and gravity models [38–43]. Among the latter models the two-step

floating catchment area method (2FSCA) stands out [44,45]. However, the adoption of an equal catchment size was criticized for the lack of nuances in interpreting the effect of decreasing distance [43] and to adapt to the different travel environments where health search behaviors take place, the enhanced two-step floating capture model (ESFCA) was proposed [46]. Nonetheless, in these latter two methods, there is the oversight of regional competition [7]. In fact, it is also known as "intervention opportunities" in the language of spatial interaction models [47], as they limit search behaviors, as in many cases, the patient can be treated outside a particular administrative unit [48,49], to minimize this defect the three-step floating catchment area method (3SFCA) was developed [50–52]. Regarding the choice of 2FSCA versus 3FSCA, it is a necessary combination of both a distance-decay function and variable catchment size function for the 2SFCA to appropriately measure healthcare access across all geographical regions [53,54]. The 3SFCA method is based on a more reasonable assumption of healthcare demand for medical services [51]. Suppose it is considered the chance that a patient could be treated in two neighboring countries. In this scenario, the boundary line would be considered non-existent to serve the population regardless of the country in which it resided. To seek collaboration between both countries to seek a better health service to the population. 3FSCA assumes that a population's healthcare demand for a medical site is influenced by the availability of other nearby medical sites. Indeed, it assigns a travel-time-based competition weight for each pair of population-medical sites in addition to the E2SFCA methodology.

In fact, when we aggregate the population within the overlapping catchment areas of multiple facilities, the original 2SFCA framework leads to double-counting of the population that tends to increase the level of demand in the healthcare system [55]. Various solutions to the demand and level of service increase have been proposed, including selecting weights based on a travel impedance function in the 3SFCA method [55].

The current work intends to elaborate a viable framework to measure spatial accessibility for the resident population in Spain and Portugal regarding hemodynamics rooms. Through the present study, the municipalities have been classified according to their accessibility degree. Therefore, it was allowed to determine which are the ones that show inadequate accessibility levels.

As for the spatial heterogeneity of border areas, by using an adjusted spatial access index, the 3SFCA method indicates strong potential for identifying health professional shortage areas [51]. In this regard, in the borderline region between Spain and Portugal, there is a shortage of hospitals, and as a consequence, there is a shortage in the sanitary services offered. Another point to consider is that the opening hours are 24 h a day. Therefore, for every hemodynamic room, the perception of quality care was taken into account by the doctors who finally decide where the patients will be treated.

In order to carry out such a study, two scenarios will be put forward: (1) the patients could be treated only in hospitals with hemodynamics rooms in their own country; (2) the patients also be treated in hospitals located outside their country. In this regard, the difference between these scenarios will identify the municipalities that improve their accessibility patterns and quantify how people's living standards continue to increase regarding health services.

#### **2. Materials and Methods**

Based on official information and the application of the spatial interaction model 3SFCA, the proposed objective could be met. In this regard, the data used comes from publications made by official institutions. Although, they must be differentiated according to the sources used for each of the countries. In the case of Spain, these sources are the Official Road Map of 2021 of the Ministry of Development, the National Cartographic Base at a scale of 1:200,000 (BCN200) of the National Geographic Institute (IGN), the revision of the Municipal Population Register of 2020 of the National Institute of Statistics (INE) and the Minimum Basic Set of Hospital data (CMBD) of 2014, the National Catalog of

Hospitals (CNH) of 2020 prepared by the Ministry of Health, the Ministry of Social Rights and Agenda 2030 and the Ministry of Consumer Affairs.

In the case of Portugal, the data comes from the web portal relating to the Portuguese infrastructure network, the National Geographic Information System (SNIG) and the National Territorial Information System, the PORDATA web portal, and the set of Health Centers in 2012 from the National Institute of Statistics (INE). Likewise, to obtain the hemodynamic room number of each hospital, it was necessary to have Computer-assisted telephone interviewing (CATI) and computer-assisted web interview (CAWI), based on electronic surveys sent to respondents before.

However, there is a common source of information for the two countries, the land use registered by the Corine Land Cover program for the year 2018, to determine the uses of the continuous urban fabric land designated with code 111 serve to determine urban areas.

Likewise, the development of all the tasks and calculations performed was carried out using the R statistical package (created by Ross Ihaka and Robert Gentleman, in Auckland, New Zealand) and the ArcGIS 10.8.1 application (ESRI, Redlands, CA, USA) and its network analysis tool, Network Analyst.

In fact, the tasks carried out are differentiated by five essential phases. Initially, the design of the base cartography, continuing with the determination of the floating catchment areas of the hospitals equipped with some hemodynamic room and continues with the obtaining of the thematic cartography that shows the degree of health coverage for each municipality in two differentiated scenarios and ends with the comparative analysis of the alphanumeric information. In one scenario, the patient can only be treated in hospitals located in the same country where he suffered a myocardial infarction—in the second scenario, taking into account that the patient can also be treated in hospitals in the neighboring country (Figure 1).

**Figure 1.** Workflow.

#### *2.1. Developing Cartography*

The cartography used is vector and is composed of four layers in Shapefile (shp.) format. Likewise, it has been used in the European Terrestrial Reference System 1989 (ETRS89) in spindle 30.

Firstly, the road network is modeled using graphic entities linear to the entire road network of peninsular highways (Figure 2). The topology generated for this layer is of the arc-node type, on which the impedance is determined in minutes as the time it takes a vehicle to travel each of the network sections. This happens once each section of the road had associated the maximum speed allowed and the distance to travel.

**Figure 2.** Roads in the Iberian Peninsula.

The second layer of information corresponds to the urban area of cities and towns (Figure 3). Polygonal graphic entities represent these, and the resident population in each one of them composes the associated alphanumeric information.

**Figure 3.** Continuous urban fabric in the Iberian Peninsula.

The third layer of information evokes the municipal capitals represented by dots with the resident population as associated alphanumeric information (Figure 4).

**Figure 4.** Municipal capitals classified by population.

The last layer represents the cities with hospitals that contain some hemodynamic rooms (Figure 5). Precisely the number of hemodynamic rooms is the alphanumeric information associated with them. The graphic entity that represents them is the centroids of the municipal capitals that contain some hemodynamic room. These were determined by selecting those municipal capitals with a public or subsidized hospital equipped with a hemodynamic room. In this regard, Figure 5 shows a symbol in the cities with hospitals where there are hemodynamics rooms. However, this symbol is also proportional according to the number of hemodynamics rooms in the hospitals.

**Figure 5.** Hospitals classified by number of hemodynamics rooms.

#### *2.2. Origin-Destination Time Matrix*

The time from the emergency notification to the balloon implantation in the PCA should not exceed 90 min according to the medical standards [56–60]. Therefore, since the patient suffers acute myocardial infarction, a hemodynamic room is crucial to the treatment in a hospital. In fact, this determines physicians' behavior when selecting the hospital where the patient is to be transferred. As a consequence, a threshold time of 90 min was established in the analysis.

In this regard, the transfer time of the patient to the hospital was determined as the travel time between the municipal capitals that require a PCA and the cities that have a hospital with at least one hemodynamic room. Besides, by calculating the inter-urban travel time between cities or towns, the intra-urban times it takes to traverse the different urban environments are estimated, from the municipality of origin where the patient suffers the myocardial infarction to different hospitals that have a hemodynamic room. Precisely, the population and the urban area are used to estimate these intra-urban times, since in this methodology they are estimated based on the population density of urban areas, through a linear adjustment that gives a maximum of 80 km/h to the areas with the lowest population density and a minimum of 20 km/ha in the most densely populated areas [59]. In this way, the sum of the inter-urban time plus the intra-urban time determines the total time of the journey between the municipal capitals with some patients due to myocardial infarction and the urban centers with a hospital hemodynamic room.

#### *2.3. Defining Hospital Catchment Areas*

The relationship between supply and demand for healthcare resources is analyzed by applying the spatial interaction model called 3-step floating catchment area (3SFCA). This model is based on a logical conjecture of the demand for health care [10]. Because it assumes that the population that demands health care from a particular place is influenced by the availability of different nearby sites where medical care is offered, specifically by the time of separation and the services offered. Conceptually, the model assigns a travel-time-based competition weight for each population-medical site pair in addition to the methodology outlined in an enhanced two-step floating catchment area (E2SFCA). This weight is then used to calculate the demand of service sites, thereby minimizing the overestimation [58]. The method is implemented in three steps:

*Step 1*: Determine the catchment of a population location *i* based on a 90-min driving zone. A person can suffer a myocardial infarction and where all the services available within the catchment are sought. In this case, all those hospitals have at least one hemodynamic room. Subsequently, a Gaussian weight is assigned to each service site according to the sub-zone in which the site lies (i.e., if a service site is located within the second sub-zone, the Gaussian weight (i.e., W 2) of the sub-zone is assigned to the service site), and calculate a selection weight between each service site and *i* by:

$$G \frac{T\_{ij}}{\sum\_{k \in \{Dist(i,k) < d\_0\}} T\_{ij}} \tag{1}$$

where *Gij* is the weight of the selection between the location *i* corresponding to the capital of the municipality where a patient could suffer a myocardial infarction, and the place *j* where there is a hospital with at least one hemodynamic room, *Dist* (*i*,*j*) is the cost of the trip (minutes) from *i* to any service location *k* within the catchment. Likewise, *d*<sup>0</sup> is the size of the basin, which in this case is 90 min, taking into account that the maximum time by medical standards [56–60]. The Gaussian weights for *j* and *k* were assigned using *Tij* and *Tik*, respectively.

*Step 2*: Determine the 90-min catchment area of each service site *j* and divide the catchment into five sub-zones using the same procedure of step 1. All locations within the catchment are sought and computed the physician-to-population ratio (*R*) of *j* by:

$$R\frac{S\_j}{\sum\_{k \in D\_r} G\_{kj} P\_k \mathcal{W}\_{r\_j}}\tag{2}$$

*Sj* is the medical capacity of *j*, in this case, it corresponds to the number of hemodynamic rooms available in each hospital. Likewise, *Wr* is the impedance of the *r*-th sub-zone *Dr*, was determined through the road network that served to move the patient from each capital of each municipality to a hospital with at least one hemodynamic room. Furthermore, this impedance considers inter-urban time and intra-urban time, and *Gkj* is the selection weight between *j* and population site *k*, and *Pk* is the population size of *k*.

*Step 3*: Compute the spatial access of population site *i* by:

$$A\_i^F = \sum\_{j \in D\_r} G\_{i\bar{j}} R\_j \mathcal{W}\_r \tag{3}$$

where *Rj* is the physician-to-population ratio of *j* within the catchment, *Gij* is the selection weight between *i* and *j*, and *Wr* is the Gaussian weight of the *r*-th sub-zone *Dr*.

In the case analyzed, the 3SFCA assumes that the demand of a municipality's population is affected by the cost of traveling to the nearest health service that offers the treatment of primary percutaneous coronary angioplasty as by its travel costs to adjacent sites offering the same service. This is a logical assumption because the demand of the people for a medical site will decrease when the adjacent sites are also available since the demand of the population in some cases could exceed the supply of hemodynamic rooms that are offered. In fact, the selection weight, *Gij*, reflects this change. *Gij* equals 1 when only one medical site is available for a population site but decreases with an increasing number of available alternatives. The multiplication of *Gij*, *Pi*, and *Wij* represents the adjusted population demand of location *i* on medical site *j*.

#### *2.4. Developing of the Thematic Cartography*

All the previous methodological steps were applied, taking into account two scenarios. In the first one, patients suffering from a myocardial infarction can only be transferred to hospitals located in their country to receive a PCA. Second, patients suffering from a myocardial infarction can be transferred to hospitals in their country or to a neighboring country. For this reason, two thematic maps were obtained.

From the results obtained in analyzing the accessibility to the PCA service, five classes were established through equal intervals, considering the maximum value obtained in both scenarios. In this way, they remained constant in both scenarios, allowing the same classification of the values corresponding to the capacity map to be made at the PCA service. Thus, it is possible to compare both thematic maps and, consequently, both scenarios, identifying and locating those municipalities that suffer the greatest and least variation if the patient could have access to health services in the neighboring country.

#### *2.5. Comparative Analysis of Alphanumeric Information*

Comparing the alphanumeric information associated with each of the municipalities in the two analyzed scenarios and the results obtained after applying the 3SFCA allows quantitatively determining the municipalities and the population residing in them according to their access coverage they offer hospitals that have a hemodynamic room.

In this regard, the execution of Structured Query Language (SQL) selected the data corresponding to the municipalities classified according to the established levels of health coverage. From this selection, the relative distribution of health coverage concerning the resident population in the analyzed municipalities and the accumulated percentage of the population could be captured, according to the capacity of PCA health services, by the municipality for inhabitants to access to any country or capacity of PCA health services, by the municipality for inhabitants that could only access they own country. Without ambiguity, it is possible to determine which of the scenarios show greater inequality in access to health services.

#### **3. Results**

The thematic maps represent the health coverage for each municipality's inhabitants in the two scenarios proposed. For this reason, the first map represents the health coverage of each municipality if the patient can only be transferred to hospitals located in the same country where he suffered a heart attack (Figure 6). The second map shows the health coverage of each municipality if the patient can also be transferred to hospitals in the neighboring country; that is, if the patient suffers a heart attack in Spain, they could also be transferred to Portugal and vice versa (Figure 7).

**Figure 6.** Capacity of health services to make a PCA, by the municipality, if patients that could only access hospitals located in the same country where they suffered a myocardial infarction and are not entitled to be treated in the other country.

The thematic map in Figure 6 shows the situation of each of the hospitals in the Iberian Peninsula that have at least one hemodynamic room to perform a PCA and the health coverage in the municipalities if the patient can only be transferred to hospitals that are in the same country where he suffers the myocardial infarction. Regarding the distribution of hospitals, it can be seen that the distribution in the border area between the two countries is scarce. In fact, there is only one hospital in a cross-border city. This hospital is located in the NUTS III of Badajoz, located in the southwestern part of Spain (Figures 5 and 6). Likewise, also in Spain, in the northwest region, there are three hospitals in areas close to the border located in the NUTS III of Ourense and Pontevedra (Figures 5 and 6).

On the contrary, in Portugal, there is no hospital in a cross-border city. However, there are areas close to the border, in the north in the NUTS III of Cávado, in the southern half in the NUTS III of Alentejo Central, and in the south at NUTS III in Algarve (Figures 2 and 6). This spatial distribution of hospitals reveals that both countries have developed health policies without coordination to achieve greater coverage in the cross-border area.

**Figure 7.** The municipality's capacity of PCA health services for patients that could access hospitals located in the same country where they suffered a myocardial infarction and hospitals in the neighboring country.

Regarding the health coverage observed in each of the countries, in Spain, no pattern is observed (Figure 6). Although, there is a predominance of municipalities with medium, high, or very high health coverage. Even in the cross-border area with Portugal. However, it is true that in Peninsular Spain, there are also some areas with some isolation from access to hospitals. The largest area located between the NUTS III of Soria, Guadalajara, Teruel, and Cuenca stands out. However, the cross-border area with France in the north of the NUTS III of Huesca and Lleida, between the NUTS III of Cáceres, Toledo, Badajoz, and Ciudad Real (Figures 2 and 6) are also noteworthy. In Portugal, a pattern is observed, since in the eastern part and bordering with Spain; there seems to be a low health coverage in the vast majority of municipalities (Figure 6). However, at the western end, coverage is greater, highlighting a center-periphery model around Lisbon. Therefore, Portugal seems to have developed a distribution of health resources in the coastal areas in the northern half and the southern half centered on the NUTS III of Grande Lisboa, south of the NUTS III West and in the NUTS III of Algarve, existing a great contrast between areas with optimal health coverage and those that practically suffer from inadequate coverage (Figures 2 and 6). Precisely, the comparison of this health coverage in the cross-border area indicates that Portugal would benefit more if there were a common policy between both countries, since it has a greater extension of territory with little health coverage, and Spain has adequate health coverage in this cross-border area.

The thematic map in Figure 7 shows, as in Figure 5, the same distribution of hospitals. However, in this case (Figure 7), the patient who suffers a myocardial infarction can also be transferred to hospitals in the neighboring country. As might be expected, the variation in health coverage occurs in the border area and not in the rest of the territory of both countries. However, the comparison of the two scenarios proposed (Figures 6 and 7) in the cross-border area shows that the effects of the variation in health coverage are different in both countries.

In this regard, in Spain, the NUTS III of Ourense (Figures 2 and 7) improves the health coverage of most of the municipalities located in the south of this NUTS III. In Portugal, in the NUTS III of Minho-Lima, some municipalities have low or medium health coverage (Figure 6) to have very high coverage (Figure 7). Possibly, because the accessibility to the hospitals located in the NUTS III of Pontevedra and Ourense is adequate, possibly made possible by an optimal network of roads.

Likewise, in the northern part of the country and in the eastern direction, several municipalities within Tras-os-Montes NUTS III improve their health coverage, going from very low (Figures 2 and 6) to low (Figures 2 and 7). In the case of Spain, some municipalities located to the south within the NUTS III of Ourense improve. Therefore, it is shown that those patients in the Portuguese municipalities located in the northwestern part of the hospital located in the NUTS III of Alto Trás-os-Montes, if they could be transferred to Spanish hospitals, they would have the same health coverage as the Portuguese municipalities closest to the hospital located in Alto Trás-os-Montes. The road network to access Spanish hospitals could be more adequate than the road network to access the hospital located in the NUTS III Alto Trás-os-Montes located in the same country (Figures 2 and 7).

Furthermore, in the NUTS III of Alentejo Central and Alto Alentejo, the health coverage of some municipalities in Portugal improves. Possibly, due to the proximity in access time to the Badajoz hospital, since the road network allows an access time consistently below 90 min in the municipalities indicated above and, also, because this Spanish hospital has a greater number of hemodynamics rooms. On the contrary, in Spain, the variation in health coverage is inexistent. Therefore, there is no improvement in this coverage, even with the possibility of transferring the patient to Portuguese hospitals—possibly, because the access time to the hospital located in Central Alentejo does not compensate for treating the patient when in the hospital located in the NUTS III of Badajoz (Spain) there is a greater number of hemodynamic rooms.

Finally, in the southern border area, the effect produced by the improvement of health coverage in both countries is non-existent. Possibly, each country has adequately endowed that part of the territory and transferring the patient to the neighboring country is not appropriate. Due to the location of the hospitals and the resources available in them, and because of the road network to reach them. In other words, in this case, contrary to what happened in the cross-border northern half, the attractiveness produced by the healthcare resources offered in the neighboring country does not overcome the inconvenience of having to transfer the patient and travel more kilometers to reach them to the hospital.

Also, from the thematic maps, the number of municipalities and the resident population were represented in percentages grouped according to the five levels of health coverage (very low, low, medium, high, and very high) previously used in the thematic maps (Figures 8 and 9).

**Figure 8.** Number of municipalities and population for patients that could only access hospitals located in the same country where they suffered a myocardial infarction.

**Figure 9.** Number of municipalities and population for patients that could access hospitals located in the same country where they suffered a myocardial infarction and hospitals in the neighboring country.

Figure 8 shows these percentages if the patient who suffers a myocardial infarction can only be transferred to a hospital in the same country where he suffered said infarction. Thus, it stands out that, although only 14% of the municipalities have very high health coverage, they host three-quarters of the population. Therefore, it can be stated that a scarce 14% of the municipalities host 75% of the population and have excellent health coverage in the Iberian Peninsula. However, there are also municipalities where there are health coverage problems. On the contrary, 11% of the municipalities with a low population of 3% have health coverage problems.

Consequently, it can be established that the most unpopulated municipalities are those with the greatest health coverage problems. In this sense, it stands out that the population increases progressively from the levels with the worst health coverage to those with the highest health coverage, from 3% to 75%. Therefore, the planning of health services has obeyed, among other criteria, the fundamental criterion of the existing population in each country's regions.

In order to be able to compare health coverage in the two scenarios, Figure 9 was made using the same criteria as in Figure 8. However, in this case, assuming that the patient could be transferred well to hospitals located in the same country where he suffered the heart attack of the myocardium or to a hospital located in the neighboring country. Nevertheless, the comparison of both figures (Figures 8 and 9) allows us to affirm that the pattern is maintained in terms of the health coverage offered. Most of the population, 76%, would continue to have better health coverage and a small population with poor health coverage. Even the same trend is observed in terms of population increase as the level of accessibility increases.

Nonetheless, the comparison of both scenarios (Figures 8 and 9) allows us to observe some patterns. In the first place, at the lowest levels of health coverage (very low and low), the number of municipalities and the population decreased slightly. On the contrary, the number of municipalities and the population with the best levels of health coverage increase (medium, high, and very high). Therefore, it seems that municipalities and their population are being removed from specific sanitary isolation. However, there are still municipalities and populations with little health coverage.

Figure 10 represents the percentage of the accumulated population in percentage values according to the percentage of accumulated health coverage. For this reason, the blue line evokes an equitable ideal distribution of health coverage for the realization of PCA for the entire population of the Iberian Peninsula. Likewise, the red line represents health coverage for the population, considering that the patient can also be transferred to a hospital located in the neighboring country where they have suffered the myocardial infarction. Furthermore, the green line represents the capacity of the health service if the patient can only be treated in a hospital located in the same country where he suffered this myocardial infarction. Thereby, the curve representing the possibility that the patient can be transferred to hospitals located in both countries is closer to the curve that represents the ideal equitable distribution. As a consequence, it can be affirmed that health coverage for PCA would be more equitably distributed if a patient suffering from myocardial infarction can be treated in a hospital independent of the country where they came from, simply taking into account the access time and the number of hemodynamic rooms at the different hospitals.

**Figure 10.** Comparison analysis through a Lorenz curve.

#### **4. Discussion**

Policies on health infrastructure between two countries that are part of the European Union should be coordinated. However, this study shows that, to some extent, this is not the case between Spain and Portugal.

The border region is made up of mostly sparsely inhabited municipalities. When it comes to health matters as a primary right of citizens, in both countries, the location of hospitals has been prioritized in those more inhabited places, or at least the establishment of those hospitals better equipped with high-tech sanitary equipment, such as hemodynamic rooms to apply a PCA.

However, greater health cooperation between the two would achieve a more equitable distribution of health coverage for all the inhabitants of the Iberian Peninsula. Although Portugal could benefit more, there would also be inhabitants of Spanish border municipalities that would benefit. Likewise, it must be taken into account that both in the border area and the interior of both countries, considering the two scenarios analyzed, there are areas with little health coverage. One would expect to apply specific mitigating measures to alleviate this possible sanitary isolation of high-tech sanitary resources on these territories.

Additionally, even if current health collaborative environments are now populated mainly of a great diversity of cooperation alternatives, the fact is that we still miss the use of specific tools providing essential and effective strategies regarding the use of medical facilities located along the border. However, as Morales et al. [48] mentioned, communication links enable information collection that can be accessed and/or used by shared infrastructures and services in these territories. Mainly, in medical and clinical environments, resource and service sharing can promote countless potential benefits, supporting very often territories in which medical specialists and infrastructures are very often reduced or limited, enhancing collaboratively gathering vital health opportunities for patients living

in these cross-border territories—which are the territories inserted near the borderland, in this case, along the Spanish-Portuguese border.

This fact is increasingly vital in a scenario in which most of the renew equipment and/or construction processes of health infrastructures are supported and financed by European Community funds, that should benefit all the European populations regardless of existing frontiers, bearing in mind the objective to grant better health care for every citizen.

#### **5. Conclusions**

Through the present investigation, a series of conclusions were reached, taking into account the access time and the number of hemodynamic rooms to the different hospitals.

Firstly, Spain and Portugal have developed health policies without consideration of the status in the other neighboring country—as shown by the poor distribution of hospitals in the cross-border area—the territories located near the Portuguese-Spanish borderland. However, it is also true that it has been shown that in Portugal, there is a greater concentration of population in the coastal area; this country may require a greater concentration of hospitals here.

Secondly, Portugal would benefit more from developing greater coordination and cooperation in health policy between the two countries—once this country has a more extensive territory where health coverage is low or very low. On the contrary, Spain has a higher level of adequate health coverage in most cross-border areas.

Thirdly, the variation in health coverage that would occur if the patient could also be treated at a hospital in the neighboring country where he suffered the myocardial infarction shows a different effect in both countries. In Portugal, in most of the affected municipalities, they could achieve better health coverage, in some cases equating them to the municipalities closest to hospitals in Portugal. However, in Spain, which already has adequate health coverage, the number of municipalities where better health coverage would be produced is scarce.

Both countries show excellent health coverage for most of the population, as 75% of the population has more than optimal health coverage. Therefore, both have followed a health policy in favor of those most populated municipalities. However, we must not forget that there is still a significant part of the population that has health coverage problems, and for these, it would be necessary to try to apply mitigating measures.

In this regard, the collected data enable us to put forward unique ideas considering that health departments and facilities located along the Portuguese-Spanish border can meet public health needs within their common jurisdictions as long as we enable border regions and their complex environment to be considered as a single European region, regardless of national boundaries. In fact, resource sharing across jurisdictions is a critical opportunity to cross-border regions enabling these territories, sometimes in disadvantaged positions to improve competitiveness, gain effectiveness and increase efficiency.

Additionally, the collected data enabled us to conclude that cross-border health resource sharing constitutes a viable and desirable process to overcome fundamental challenges within scenarios of increasing constraints posed by restricted budgets. Indeed, even if this is not the only reason for cross-border cooperation, limited health resources and budgets associated with low-density territories as the ones characteristic from the majority of the Portuguese-Spanish border, should consider, augmenting resource sharing not only to address emerging challenges but to grant better health services to local populations.

Lastly, although the pattern of health coverage is not broken in the two compared scenarios, it can be established that it is more favorable for the whole of the Iberian Peninsula in the second scenario. In this sense, the municipalities and the population fall for the most precarious levels of health coverage, increasing these in better levels of health coverage. Likewise, health coverage would be distributed more equitably regardless of the country where the patient suffers the myocardial infarction.

The originality and relevance of this study should be emphasized on its findings once it is possible to find an issue for such a relevant issue as is the case of the emergency

health problems of the Portuguese-Spanish borderland populations. Nevertheless, if this study outcome gives us a significant contribution to this thematic field of crossborder cooperation, some improvements and future research lines persist. Among the various research lines that persist, we can use different testing tools—i.e., use a Gini coefficient instead of a Lorenz curve. Therefore, the results from the analysis would be more informative; consequently, more insights could be obtained.

**Author Contributions:** Conceptualization, José Manuel Naranjo Gómez and Rui Alexandre Castanho; methodology, José Manuel Naranjo Gómez; software, José Manuel Naranjo Gómez; validation, José Cabezas Fer-nández., Rui Alexandre Castanho and Luís Loures; formal analysis, José Manuel Naranjo Gómez; investigation, José Manuel Naranjo Gómez and Rui Alexandre Castanho; resources, José Cabezas Fernández; data curation, José Manuel Naranjo Gómez; writing—original draft preparation, José Manuel Naranjo Gómez; writing—review and editing, Rui Alexandre Castanho; visualization, Luís Loures; supervision, José Cabezas Fernández; project administration, Rui Alexandre Castanho; funding acquisition, José Cabezas Fernández, Rui Alexandre Castanho and Luís Loures. All authors have read and agreed to the published version of the manuscript.

**Funding:** This publication has been made possible thanks to funding granted by the Consejería de Economía, Ciencia y Agenda Digital from Junta de Extremadura and by the European Regional Development Fund of the European Union through the reference grants GR18052 and GR18054.The project is funded under the program of the Minister of Science and Higher Education titled "Regional Initiative of Excellence" in 2019–2022, project number 018/RID/2018/19, the amount of funding PLN 10 788 423,16. Besides, the authors would like to acknowledge the financial support of the National Funds Fundação para a Ciência e a Tecnologia, I.P. (Portuguese Foundation for Science and Technology) by the project UIDB/05064/2020 (VALORIZA—Research Centre for Endogenous Resource Valorization).

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** The data presented in this study are openly available. Also, it is possible to contact one of the study authors.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


### *Article* **Exploring Equity in Healthcare Services: Spatial Accessibility Changes during Subway Expansion**

**Maohua Liu, Siqi Luo and Xishihui Du \***

School of Transportation Engineering, Shenyang Jianzhu University, Shenyang 110168, China; cemhliu@sjzu.edu.cn (M.L.); yogalinrose@163.com (S.L.)

**\*** Correspondence: daisy\_duxi@126.com

**Abstract:** The unequal allocation of healthcare resources raises many fundamental problems, one of which is how to address inequity in population health. This paper focuses on disparities in public transport healthcare accessibility, with a special focus on an expanding subway system. Based on a vulnerability index, including factors that are likely to limit healthcare opportunities, a twostep floating catchment area method was used to assess the distribution of supply and demand for healthcare. Quantity, quality, and walking distance accessibility were aggregated into hexagonal grids. The Theil index was used to measure inequity and understand the influence of subways on spatial disparities in healthcare accessibility. The ongoing construction of the subway has heterogeneous impacts on healthcare accessibility for different parts of the city and exacerbates spatial inequity in many areas. In an environment where people in peri-urban areas are excluded from healthcare access because of low subway coverage, the results suggest that the potential for subways to address inaccessibility is limited. The findings highlight the requirement of efficient public transport services and are relevant to researchers, planners, and policymakers aiming to improve accessibility to healthcare, especially for populations who dwell in winter cities.

**Keywords:** geospatial health; spatial disparities; accessibility; GIS; subway expansion; public transport network

#### **1. Introduction**

Equity matters for every social group because it raises opportunities and supports the rights that should be available to every individual within a population. If equity among the population is high, the society benefits overall [1]. However, many public transport (PT) systems do not provide adequate services for citizens to easily access public resources or to meet complex travel needs. For example, in many cases, low-income subdistricts are more heavily dependent on PT [2], and a simple PT system may not provide adequate access to groups with high service requirements, such as complex journeys [3]. In high demand regions, inadequate PT may limit access to resources and opportunities, making them more susceptible to social and economic marginalization [4,5]. Equity has attracted considerable attention in relation to public resources and urban infrastructure because a mobility gap often exists between PT availability and population demand. This gap has brought to prominence two key research topics addressing the spatial equity of public resources. First, accessibility based on sociodemographic attributes (e.g., age, gender, race, income), which can highlight inequities in individuals' access to public resources [6–8]; and secondly, in response to this, analysis of locations through spatial optimization of both facilities and transport networks [9,10].

Spatial equity analysis focuses on differences in the services used by different regions or social groups from the perspective of supply and demand; and is, to some extent, an extension of the concept of accessibility [11]. Therefore, quantifying spatial accessibility is an important foundation for measuring spatial equity, hence assessing social equity [12,13]. In the field of health and transportation, measuring accessibility plays an important role in

**Citation:** Liu, M.; Luo, S.; Du, X. Exploring Equity in Healthcare Services: Spatial Accessibility Changes during Subway Expansion. *ISPRS Int. J. Geo-Inf.* **2021**, *10*, 439. https://doi.org/10.3390/ijgi10070439

Academic Editors: Fazlay S. Faruque and Wolfgang Kainz

Received: 29 April 2021 Accepted: 24 June 2021 Published: 27 June 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

comprehensively evaluating the equity of service distribution within a region [14,15]. Wee and Geurs [16] proposed that lack of access to opportunities is the most important indicator of transport-related inequity. Measures of access include both the availability of an activity (such as work, education, shopping, healthcare, or recreation) and the ease of access to the location of the activity from a given origin, usually a residential location. Accessibility is mainly influenced by two factors: the balance of supply and demand relative to origin and destination points, and the suitability for purpose of the transportation network. Urban transportation systems are gradually upgraded, and reconfigured, in response to ongoing rapid urbanization and increasingly complex distributions of urban functions, service resources, and diverse people's demands. Estimates of accessibility depend on the factors themselves but are also substantially influenced by the choice of accessibility measure. Four measures are commonly used to evaluate place-based accessibility: gravity-based accessibility, cumulative opportunity accessibility, utility-based accessibility, and emerging measures based on real-time individual data [17]. Gravity-based measures utilize a distance/time decay function to normalize the cost of travel between the origin and destination. This approach follows the gravity model's assumption that the interactions between activities are directly proportional to their size and inversely proportional to the cost of traveling between them [18]; however, the cost decay is uncertain for every group or individual [19]. The cumulative opportunity measure assumes that individuals will utilize the opportunities nearest to them, and more nearby opportunities translate into more choices for individuals. There is no limit to the capacity of these opportunities [20]. However, people often compete for the same opportunities and, in the case of employment, one job can only be taken by a single person. Thus, the assumption that more opportunities will translate to more choices, without considering the potential demand and hence competition for those opportunities, can be misleading [5]. Utility-based measures estimate the value of opportunities based on the assumption that users/consumers of a transport system seek to maximize the utility of their behavioral choices. This is a cross-disciplinary approach utilizing economic, social, land use, and transport data, and is still considered an emerging method that requires substantial research and development [21,22]. The final category is real-time-based accessibility measurements. Influenced by temporal geography, some scholars have researched spatiotemporal accessibility available to individuals, which is expected to accord with social reality [4,23]. They focus on the impact of the scale of public service facilities, transportation mode choices, and real demands on the equity of access to those facilities [10,24,25]. The above four measures have in common either distance/time cost or approach opportunities as indicators. However, accessibility estimation is likely to be influenced by the measurement method, which may limit the assessed access possibilities through limits to distance or accessibility opportunities. In general, commuters are usually concerned about commuting time and distance, which can be measured by gravity-based accessibility. When shopping, consumers are more interested in the variety of goods, their quality, and their prices than by distance [26,27]. In healthcare or education, users are sensitive to the quality of public services in addition to the travel options and costs [28–30]. Therefore, discussing accessibility from a single perspective is only a partial solution, and a customized measurement method should be developed for different behavioral activities. Therefore, in this study, we propose measuring accessibility to healthcare services from diverse perspectives, based on citizens' concerns and interests, and the consequences for social equity.

As important public facilities, a reasonable spatial distribution of healthcare services is influential in people's livelihood and security. Many studies have demonstrated that poor access to healthcare services contributes to lower levels of service utilization, which in turn leads to poorer health outcomes [31–33]. The provision of adequate and equitable healthcare access across the whole population has become a concern for governments and societies [34–36]. Public healthcare plays an important role in meeting the health needs of the population, and seniors especially are considered to be among the most vulnerable groups in the population [37]. Although the elderly are the main demand group for healthcare, their overall health status in China is poor. Nearly 180 million elderly people suffer from chronic diseases, and the proportion of those suffering from one or more chronic diseases is as high as 75%. The number of people aged 60 years and over in the city of Shenyang is estimated to have reached 2.019 million by the end of 2020, accounting for 26.52% of the total population. Consequently, it is vital for Shenyang to make provisions for equal access to healthcare services. Especially in winter, large numbers of people in Shenyang (a designated "winter city") face difficulties walking outdoors, and residents rely on PT for daily travel. Based on data from Shenyang's comprehensive traffic survey conducted in the downtown area in 2017, 32.8% of all journeys were made by PT. The rapid development of China's subway systems has provided a new transportation alternative to citizens in winter cities and has additionally helped mitigate multiple urban health and environmental challenges such as congestion, traffic injuries, air pollution, greenhouse emissions, and noise. Owing to its cost and speed advantages, the subway has become the preferred mode of transportation for most residents, especially when individuals seek healthcare alone. While subways have been undergoing revitalization as one of the country's primary PT modes, inequity in healthcare services has remained a major concern for health planners and policymakers.

We propose an approach for measuring healthcare accessibility from the perspective of subway expansion. We use a vulnerability index and accessibility measures to compare healthcare access equity and so the influence of subway expansion. For this purpose, vulnerability factors that are likely to limit healthcare opportunities have been proposed, including income, recent immigration, population age distribution, and physical conditions. Based on the obtained population vulnerability and healthcare data, the distribution of the supply–demand balance of facilities was visualized using a two-step floating catchment area (2SFCA) method. Furthermore, a comprehensive accessibility measure was used relating quantity, quality, and walking distance in the context of subway upgrading. Finally, the Theil index was used to measure the equity of healthcare resources. The proposed method will provide a more nuanced understanding of spatial disparities in healthcare accessibility, and the results of this empirical study will offer new insights into the ways in which variations in PT influence healthcare accessibility. Officials in the fields of public health and planning can reduce local disparities by designing targeted interventions.

#### **2. Materials and Methods**

#### *2.1. Study Area*

Shenyang is the capital of Liaoning Province and is an important city in northeastern China. Our study area (Figure 1) includes the central urban area based on the third ring road of Shenyang City (Figure 1c), and incorporates nine districts: Heping, Shenhe, Dadong, Huanggu, Tiexi, Sujiatun, Hunnan, Yuhong, and Shenbei New District, with a total land area of 12,860 km2. The Hun River runs through central Shenyang from east to west and divides the central area into two parts. According to demographic data from the Shenyang Statistics Bureau (2019), the number of permanent residents in Shenyang was 8.32 million in 2019. However, the populations' requirements for medical care are addressed by a relatively small number of general hospitals, with only 181 in Shenyang's central urban area. According to Shenyang's comprehensive traffic survey, which surveyed the pathways used by residents to access social resources such as healthcare services, the majority chose to use PT [38]. The PT system is known as the Shenyang Rail Transit, and included 478 bus lines and four built subway lines at the end of 2020. Its first subway line commenced operation on 27 September 2010, making it the first in northeastern China. By April 2020, there were four lines in operation (metro lines 1, 2, 9, and 10) with 92 stations and 117.06 km of operating distance, covering the central urban area. Subway lines 3 and 4, as well as extensions to lines 1 and 2, are due to be completed by the end of 2025. By then, Shenyang's main and sub-cities will be connected by the subway network, promoting interaction between subdistricts.

**Figure 1.** Spatial context of the central urban core of Shenyang: (**a**) location of Shenyang in China; (**b**) administrative boundaries in Shenyang; (**c**) overview of the study area.

#### *2.2. Data Sources and Preprocessing*

Generally, healthcare accessibility at any given location depends on three components: the capacity of the healthcare services (e.g., the number of physicians or beds), the potential demand for healthcare services (i.e., population), and transport network performance (i.e., travel impedance from locations with demand for healthcare services). The data used in this study consisted of three categories: population-based demand, healthcare services, and PT data. Population-demand data were extracted from the Shenyang Statistical Yearbook provided by the Shenyang Bureau of Statistics in 2019 and included data related to social identities (e.g., recent immigrants), health inequalities (e.g., elderly population and mortality), and socioeconomic determinants (e.g., tax revenue). Furthermore, healthcare services data (including geographical locations, hospital rank, and the number of beds) were mainly collected from an online medical service website called 99 Hospital Library (https://yyk.99.com.cn/, accessed on 25 December 2019). The hospitals selected in this study were 181 public general hospitals with high-quality medical services, which are economical and more advanced medical equipment; these hospitals are more likely to be included in the social insurance system than private hospitals, specialized hospitals, and other medical institutions [39]. Considering the limitations of adding large-scale public hospitals for decades, public hospitals in 2019 were selected to investigate accessibility. In addition, data on PT connections between various origins and destinations were derived

from the AutoNavi Open Platform (https://lbs. amap.com, accessed on 10 January 2020) by implementing Python-based web crawling technology, and were supplemented by material from the Shenyang Metro website (http://www.symtc.com/, accessed on 15 January 2020). The program used basic information on the 478 bus lines and six subway lines in the central urban city of Shenyang.

#### *2.3. Methodology*

2.3.1. Supply and Demand of Subdistricts for Healthcare Services

The vulnerability index, for healthcare facilities, of every subdistrict was calculated using weighted factors, which vary by region. As this study focused on accessibility via PT, the vulnerability index draws on characteristics that increase the likelihood of an individual's demand for PT. Following the study by Boisjoly et al. [40] and the characteristics of the study area, we selected the following indicators as the relevant variables for the vulnerability index: (i) tax value (I), (ii) number of elderly (U), (iii) number of immigrants (M), and (iv) mortality (N). The final vulnerability index is given by Equation (1), where *Zx* represents the z-score of the variable *X*.

$$V = -Z\_I + Z\_{\!\!\!I} + Z\_M + Z\_N \tag{1}$$

Subsequently, the 2SFCA method was used to analyze the supply and demand of healthcare services, which is essentially a summation of the service-to-demand rate at residential locations [41]. The demand–supply ratio *Ri* is given by Equation (2) as follows:

$$R\_{\bar{i}} = \sum\_{j \in \{d\_{\bar{j}} \le d\_0\}} T\_{\bar{j}} = \sum\_{j \in \{d\_{\bar{j}} \le d\_0\}} \frac{S\_{\bar{j}}}{\sum\_{k \in \{d\_{\bar{i}} \le d\_0\}} V\_k}, \; d\_0 = 1.5 \text{ km} \tag{2}$$

where *V* is the vulnerability index at a residential location *i* that can reach a given service by PT, *j* denotes a healthcare service, *Sj* represents the capacity of each healthcare service *j* (number of beds), and *di* and *dj* are the shortest walking distance from residential location *i* and healthcare service *j* to a PT station, and *d*<sup>0</sup> denotes a walking threshold that represents the distance from each residential location or healthcare service to the nearest PT stop. The application of Equation (2) involves two steps: the first step determines *Tj* as the service capacity of each bed, and the second step calculates the summation of the serviceto-demand rate for each subdistrict (*Ri*). The larger the value, the better the supply relative to the demand.

#### 2.3.2. Accessibility of Population Residential Location to Healthcare Services

As shown in Figure 2, the OD cost matrix and spatial connection in a geographic information system (GIS) were used to evaluate spatial accessibility for citizens in Shenyang to obtain healthcare services that find the best accessibility path from origin to destination in the transport network [42–44]. The service radius of PT stations is 0.5–1 km, and walking speed of residents is 4.5–5 km/h [45–48]. Considering the suburb residents and acceptable distance of 0.5–1.5 km for elderly to PT station in the study area, we used 1.5 km as walking threshold to analyze accessibility using OD cost matrix. Based on the 1.5 km OD's search radius, we evaluated accessibility from the perspectives of the quantity, quality, and walking distance. Quantity accessibility, *Av*, refers to the number of healthcare resources that can be accessed by PT from each residential location, which can be divided into none, low, moderate, and high accessibility at equal intervals. However, healthcare equality is a relatively comprehensive concept, which should be evaluated by the number of beds, professional physicians, nurses, and grades; according to China's healthcare services standards, healthcare conditions such as beds or professional physicians partly represent the scale and quality of medical institutions [49,50]. Therefore, the total number of healthcare beds available to individuals represents the quality accessibility (*Aw*), which can be divided into none (0), low (1–100), moderate (100–500), and high (>500) accessibility based on China's hospital classification standards. Walking to PT stations

is the main obstacle for the elderly in using PT services [51]. Hence, walking distance accessibility, *Ax*, can be represented as the total walking distance when taking PT from the origin or destination. Considering the acceptable distance to PT stations and walking speed, walking distance accessibility can be divided into none (>3 km), low (2.5–3 km), moderate (1.5–2.5 km), and high (0–1.5 km). The *Av*, *Aw*, and *Ax* are given in Equation (3).

**Figure 2.** Schematic diagram of measurement of spatial accessibility to healthcare services: (**a**) measure spatial accessibility; (**b**) the accessibility of residential location i; (**c**) aggregate accessibility measure.

Finally, residential location accessibility to hexagonal cells was integrated to minimize orientation bias from edge effects and clearly identify the differences between grids [52,53]. Through experiments, ideal hexagonal diameter was determined to be 1 km, including an average of 2.2 residential locations, thereby providing sufficient accuracy to summarize the results.

$$\begin{aligned} A\_{\upsilon} &= \sum\_{i} f \left( d\_{i}, d\_{j} \right) \\ A\_{\upsilon} &= \sum\_{i} S\_{j} f \left( d\_{i}, d\_{j} \right) & f \left( d\_{i}, d\_{j} \right) = \begin{cases} & 1 \text{ if } d\_{i} \le d\_{o}, d\_{j} \le d\_{o} \\ & 0 \text{ else} \end{cases} \\ A\_{x} &= d\_{i} + d\_{j} & d\_{0} = 1.5 \text{ km} \end{aligned} \tag{3}$$

#### 2.3.3. Equity Evaluation

To better understand both inter- and intra-regional variations in accessibility, the Theil index was adopted to identify disparities in equity. This index reflects the relationship between demand among the population living in the study area and accessibility using the PT system [54,55]. In general, the degree of relative advantage or disadvantage among groups in a population can be estimated using the Theil index (*T*), which ranges from 0 to 1 (Equation (4)). The larger the T, the greater the difference between regions, suggesting more unbalanced development. According to the equity principle, ideally all citizens in the subdistricts have equal access to healthcare services. Therefore, the ideal for planners is to minimize *T*:

$$T = \sum\_{j}^{m} \frac{V\_{j}}{V\_{\text{tot}}} \cdot \frac{A\_{j}}{\overline{A\_{\text{g}}}} \cdot \ln\left(\frac{A\_{j}}{\overline{A\_{\text{g}}}}\right) \tag{4}$$

where *m* is the number of subdistricts and their districts, *Vj* is the vulnerability index of group *j*, *Vtot* is the sum of the *m* vulnerability indexes, *Aj* is the accessibility, and *Ag* represents the average of the *Aj* values in the study area.

#### **3. Results**

#### *3.1. Spatial Distribution Characteristics of Healthcare Supply and Demand*

This section presents the results of the subdistrict supply and demand analysis for the central urban core of Shenyang. Vulnerability indicators were variable across the nine districts (Table 1). The elderly population ranged from approximately 250 thousand in Tiexi District to approximately 76 thousand in Hunnan District. The mortality rate also varied greatly, from 10.34‰ to 6.42‰. Heping District had the highest tax value, followed by Shenhe District. Interestingly, despite having the lowest tax value and the worst economic development level, Yuhong District was the most popular migration option. The high diversity of regions inevitably led to varying needs for PT. Accordingly, this section reviews the distributions of supply and demand for healthcare services in Shenyang.

**Table 1.** List of vulnerability factors and vulnerability index in each district.


As shown in Figure 3, there were several subdistricts on the edge of the central core that have the lowest supply–demand ratio. There are very few healthcare services in these areas, posing challenges to residents. On the one hand, these districts have the lowest populations in the region, but their vulnerability index is high relative to their counterparts. On the other hand, PT stations in these regions are sparsely located; therefore, residents find it harder to access healthcare services. Of the subdistricts with high vulnerability for social services, 60.2% showed lower levels of supply and demand despite having considerable geographical advantages and good PT coverage within the central urban area. Residents in the central region had a moderate level of supply–demand ratio, while their need was moderate or low. Finally, subdistricts with a higher supply–demand ratio, including the group with the highest ratio, were mainly in the northeast of the core area, which has an intensive transport network. Accordingly, the state of supply and demand was in disequilibrium in that area, and most subdistricts likely lacked the healthcare resources to match their needs. Hence, it is important to assess whether upgrading of the subway system in this area can improve access and equity.

#### *3.2. Accessibility across the Central Urban Area after Introducing Subways*

Although the supply and demand analysis provided insights into the distribution of healthcare sources and PT, it failed to elaborate on how extending subway services affects accessibility. In this section, quantity, quality, and walking distance are calculated to assess the spatial distribution of accessibility, and to compare differences following expansion of the subway.

**Figure 3.** Spatial distribution of supply–demand to healthcare services in the main urban area of Shenyang.

#### 3.2.1. Quantity Accessibility

As shown in Figure 4, quantity accessibility radially decreased with distance from the city center. However, the highest accessibility was not found in the center, but rather near the surrounding metro transfer stations. In areas influenced by metro lines 9 and 10, quantity accessibility exhibited (Figure 4b) a dual-core distribution centered on metro interchanges (metro lines 1 and 9; metro lines 1 and 10). The construction of new subway extensions is projected to clearly expand the dual-core range (Figure 4c). Zones with moderate quantity accessibility are scattered across the heartland. With subway extension the number of metro transfer stations that surround the periphery increases more obviously in the north (Huanggu and Shenhe Districts) than in the south (Hunnan District). Zones with low accessibility are mainly located outside the metro core area and are minimally influenced by the new subway building.

**Figure 4.** Spatial quantity accessibility to healthcare services at different stages of subway system development: (**a**) accessibility in 2015; (**b**) accessibility in 2020; (**c**) accessibility in 2025.

Quantity accessibility changed over time in every district (Figure 5). Residential locations with high accessibility increased from 13.89% to 32.45% over the study period, and 94.96% of the area showed improved quantity accessibility caused by changes in the subway system. As the subway gradually formed a network, the Tiexi District showed the most notable increase (19.16%) in accessibility. As a result, zones of moderate quantity accessibility were evenly distributed, except in Sujiatun District, which showed a strong

change over a decade. The percentage of moderate stage area decreased from 66.64% to 48.08%, and inaccessible zones within each district were reduced by 5.04%.

**Figure 5.** Spatial changes in residential zones with quantity accessibility ranking.

#### 3.2.2. Quality Accessibility

The spatial distribution of quality accessibility (Figure 6) is similar to quantity accessibility, and the healthcare quality available to residents increases as the subway expands. The healthcare quality in residential areas around metro line 1 is the highest, followed by successive decreases outside the core. Upgrading of the subway expands the highestquality accessibility zones to the north with metro line 3 and west with metro line 10. Notably, the southern region changes little over the study period, which indicates that the healthcare quality available to the southern population is inadequate. This outcome is likely related to the Hun River obstructing subway construction and the concentration of healthcare services in the heartland.

**Figure 6.** Spatial quality accessibility to healthcare services at different stages of subway system development: (**a**) accessibility in 2015; (**b**) accessibility in 2020; (**c**) accessibility in 2025.

Quality accessibility occurs in every district following subway upgrades (Figure 7). In particular, Heping and Shenhe Districts were influenced by metro lines 3 and 4. Because of the expanded subway, 6.19% of residential locations improved from low healthcare quality accessibility. In addition, 19.62% of residential locations experienced an increase in quality accessibility from moderate to high; most of these locations are in southern Hunnan and Huanggu Districts and are influenced by metro lines 9 and 10. In the dual-core region, the high-quality range expanded; specifically, 18.47% of residential locations newly achieved high-quality access to over the study period.

**Figure 7.** Spatial changes in residential zones with quality accessibility ranking.

#### 3.2.3. Walking Accessibility

As expected, walking accessibility increases dramatically as a result of subway expansion, and residential locations with high walking accessibility generally occurring on both sides of the subway lines (Figure 8). Viewed holistically, the high walking accessibility coverage expanded from the surroundings of metro lines 1 and 2 to the whole enclosed area of old and new metro lines. The addition of lines 9 and 10 clearly reduce walking distances in many districts, apart from Shenbei New District. However, most districts appeared to have been minimally influenced by the construction of lines 3 and 4, although a substantial reduction in walking distance was observed in Heping District (from 2767.37 m to 806.05 m).

**Figure 8.** Spatial walking accessibility to healthcare services at different stages of subway system development: (**a**) accessibility in 2015; (**b**) accessibility in 2020; (**c**) accessibility in 2025.

Overall, subway extensions increase walking accessibility (Figure 9). Most districts in the center city become highly walkable, with Dadong and Tiexi Districts showing increases owing to metro lines 9 and 10, and Heping District influenced by metro lines 3 and 4. However, much of Sujiatun District was beyond the acceptable walking range. Approximately 31.48% of residential locations showed a rise in high walkability as subway lines were extended, with 13.89% of residential locations being within the convenient walking threshold (≤1 km). Residential locations with moderate walkability only increased by 2.75%; these locations were distributed within the area enclosed by metro lines 1, 3, 9, and 10. Furthermore, 29.19% of residential locations with previously low walkability, which is far beyond the acceptable walking threshold, had improved walkability.

**Figure 9.** Spatial changes in residential zones with walking accessibility ranking.

#### *3.3. Equity Changes with Subway Line Extensions*

While clear regional differences were observed in the temporal and spatial characteristics of accessibility, it is less easy to estimate equity across subdistricts or districts. The Theil index shows the degree of inequity within regions, based on the vulnerability index and accessibility.

#### 3.3.1. Equity between Subdistricts

Although subway services have increased access to healthcare services in Shenyang, there are two ways in which inequity between regions may change. As shown in Figure 10, the impacts of subway lines 9 and 10 on equity were mixed; these lines increased the equity of quantity accessibility in 39 subdistricts, most of which were located in the extreme north or south of study area (Figure 10a). However, inequity increased for 30 subdistricts in the urban core. Further changes in equity occurred when the subway lines were further extended (Figure 10b). Equity between subdistricts clearly improved within the northeast as a result of extension of subway line 1. In contrast, inequity increased in 50 subdistricts mostly north of the Hun River. Only 12 subdistricts were not affected by building subways, mostly on the periphery.

**Figure 10.** Variations in equity of accessibility to healthcare services during development of subway system: (**a**) changes in equity of access quantity after introduction of metro lines 9 and 10; (**b**) changes in equity of access quantity after introducing metro lines 3, 4, and extension lines; (**c**) changes in equity of access quality after introducing metro lines 9 and 10; (**d**) changes in equity of access quality after introducing metro lines 3, 4, and extension lines; (**e**) changes in equity of walking access after introducing metro lines 9 and 10; (**f**) changes in equity of walking access after introducing metro lines 3, 4, and extension lines.

The influences of subway lines 9 and 10 on equity of access quality were mainly concentrated within the central strip of Shenyang (Figure 10c). Specifically, subway lines 9 and 10 improved equity within 52 subdistricts, with 15 subdistricts located in the heartland having the most obvious improvements. However, 29 subdistricts near the east and west extremes of Shenyang were little influenced by the subway lines in this regard. In addition, there were 17 scattered subdistricts in which the presence of the subway exacerbated the gap in access quality suggesting that building subways is capable exaggerating social

inequity (Figure 10d). There were 19 subdistricts far from the subway, where equity in access quality did not change, and only 34 subdistricts showed reduced disparity. Notably, access quality gaps widened within 45 subdistricts largely located in the northwest, which has a high population density and considerable healthcare needs.

Subway lines 9 and 10 reduced equity among subdistricts in terms of walkability. However, they had a positive impact on groups to the north of the Hun River, where the population needs are more diverse than in other areas (Figure 10e). Equity did not change, with subway lines 9 and 10, for 22 subdistricts located mainly in the suburbs and 11 subdistricts in the northwest were unaffected by subway building. There were 31 subdistricts with a large increase in walking distance equity, all of which were near building subways. The remaining 34 subdistricts, however, experienced aggravated walking disparity, with the largest gaps being in urban cores well supplied with PT (Figure 10f).

#### 3.3.2. Equity between Districts

As shown in Figure 11a, the addition of metro lines 9 and 10 increased inequity, as measured by the Theil index, in most districts; the variation was between 0.037 and 0.015 (except for Sujiatun District, which has no subway running through it). Furthermore, the access quantity to healthcare services in 2025 is expected to be more uneven in Yuhong District than in 2015, while the Dadong and Sujiatun Districts are the opposite. The maximum intra-district variation was in Yuhong (0.053), and the district with the least intra-district variation was Shenhe (0.035), which demonstrates that the addition of the subway has a minimal impact on quantity equity. The district with the largest inequity was the Sujiatun District (0.274), followed by Hunnan District (0.219), probably because the periphery of the urban core is sparsely populated.

**Figure 11.** Comparison of equity of healthcare services through development of subway system: (**a**) the Theil index of quantity; (**b**) the Theil index of quality; (**c**) the Theil index of walking distance.

As shown in Figure 11b, metro lines 9 and 10 had a less effect on equity of healthcare quality (the Theil index declined in most districts). However, building subways can also increase regional inequalities. The difference between the Shenhe (0.305) and Sujiatun Districts (0.274) was greatly increased by the addition of the subway. Metro lines 3 and 4 ensured accessibility improvements while also generally narrowing the gap between vulnerability groups (maximum variation was from 0.305 to 0.135 in Shenhe District). Conversely, the addition of subway services exacerbated the inequity of healthcare quality among subdistricts in Hunnan District (the Theil index changed from 0.155 to 0.244).

Finally, the disparity in overall walking access was remarkably greater than that of the other indexes (the maximum variation was 0.252–Figure 11c). Metro lines 9 and 10 decreased walking inequity in Shenhe District (from 0.451 to 0.409) and Yuhong District (from 0.210 to 0.238). However, the Theil index in the Dadong and Huanggu Districts clearly increased after adding metro lines 9 and 10. In addition, the building subway lines are likely to increase the inequity between subdistricts, suggesting that increases in quantity and quality to healthcare is accompanied by a greater walking disparity.

#### **4. Discussion**

Improving equity in access to healthcare resources using PT can potentially improve the well-being of individuals and have implications for the equalization of social resources. This study considers subway expansion in Shenyang, China as a case study to explore the ways in which subway extensions can influence accessibility and equity in healthcare facilities. When people relied on bus lines for access to healthcare resources supply and demand values in the main urban area were generally low and had a vast range. As the underground railway was built, differences in supply and demand for healthcare services were disproportionately influenced by the subway system; these findings are similar to previous research in the same context [28,56]. However, Shenyang's centralized urban structure, in relation to the locations of healthcare facilities, creates an imbalance for lowincome citizens. These conditions are endemic to many cities in China and lead to marked spatial and social health divides that translate into resource inequity and exclusion for citizens [39,57,58]. Low accessibility in peri-urban areas is a clear example, with people in peri-urban areas being excluded from healthcare options owing to the low coverage of subway lines. Our results suggest that subway expansion, on its own, has limited potential to address problems of accessibility, and its effect is only important in regions with dense PT networks. In previous studies, the influences of PT, including subways, on accessibility were multiple. The expected positive effects of PT on service access have been observed in many metropolitan areas [59]. Although PT plays a role in areas far from healthcare services along metro lines and bus routes, in some areas it may not increase access to healthcare services as much as previously thought [60,61]. Therefore, we analyzed accessibility to healthcare facilities, as a consequence of subway extension, to investigate the system's impact.

Sociodemographic diversity in each region has led to a focus on access to healthcare services. For example, in areas with a low population density and a long distance from healthcare services, people are concerned about both the walking distance to healthcare services as well as quality of the accessible services. With the addition of metro lines 9 and 10, walking distance and quality in Hunnan District clearly improved. In areas with inadequate healthcare services and unmet population needs, quantity accessibility (limited choices) is a key problem (Huanggu District). Tiexi and Shenhe Districts, which have large elderly populations, also have the highest density of healthcare services; however, the elderly have reduced walking capability. Therefore, the subway system must effectively improve walking access for areas with large elderly populations.

We suggest that analysis of the influence of a subway on healthcare accessibility should follow periods of subway extension, as well as focusing on various equities for sociodemographic diversity. Regions with unique locational advantages and highly developed transportation networks generally do not have excessive disparities in access quantity, quality, or walking distance, and typically have higher overall equalization [49,62]. However, costs are higher in areas that are far from urban centers and have larger rural populations, as these areas typically have low economic development levels and are far from hospitals [61,63]. These areas often face extreme inequalities. Therefore, the pursuit of equity in access quality is the primary requirement for such areas under limited conditions. However, the subway system has exacerbated the access quantity and quality inequalities in certain locations, such as Hunnan District, indicating that extension of a subway system does not necessarily improve social resource equity, including healthcare. The analysis confirms that, as with road infrastructure, a disproportionate concentration on high accessibility may lead to spatial inequity in transportation, as suggested by earlier studies. This explains how equity of opportunity can be influenced by the existence of PT [17,40]. Therefore, bridging

gaps in healthcare access requires more than just a subway as a mode of transportation, and PT becomes a non-obvious option to obtain the necessary access to medical assistance.

This study focused on improving the delivery of healthcare services by PT and reducing disparities; these are two key goals for planners and policymakers. On the one hand, solutions should address spatial variations in accessibility. Specifically, policies should be introduced to increase the supply capacity of healthcare services by building additional hospitals and encouraging the hiring of more doctors and nurses in suburban areas with high demand. Moreover, planning departments should introduce new PT options in these areas, as they can lead to shorter travel times for citizens when accessing healthcare services. On the other hand, targeted suggestions based on specific needs arising from population diversity should be considered by health authorities. For example, increasing geriatric hospitals in areas with high elderly needs, or new and appropriate hospital departments, may alleviate the imbalance between supply and demand. Further, planners and policymakers should be aware of the impact of subway systems on healthcare equity. Awareness of temporal variations in healthcare accessibility, using diverse measures, during PT system development gives planners and policymakers greater insight into accessibility issues. Each location can have a profile of healthcare demand and supply, hence, specific problems that require specific solutions.

It is important to acknowledge the limitations of this study, which we hope to address in the future. First, the mode of transportation used in this study was PT. Despite Shenyang's cold climate, citizens can also drive and walk for movement. Thus, the results may not completely reflect the real healthcare accessibility conditions. All available modes of travel should be considered. Second, healthcare services and population demand have changed in a decade, which should be taken into consideration. Finally, the situations in which healthcare services may be sought was incomplete; we only considered cases in which people seek healthcare services on their own. Although this is a high proportion of cases, emergency medical services should also be considered. Enrichment of the analysis to include other modes of transport and modes of hospitalization is the next step in future studies.

#### **5. Conclusions**

This study comprehensively examined spatial accessibility and equality in healthcare services in the central region of Shenyang, following an extension of its subway network. A vulnerability index was calculated to indicate the required conditions for sufficient access to healthcare services. Citizen accessibility and equity was further assessed in terms of quantity, quality, and walking distance. These factors were compared over a period of subway extensions. The results showed that upgrading the subway had spatially heterogeneous impacts on healthcare accessibility, especially walking accessibility. Construction of the subway also exacerbated spatial inequity in healthcare accessibility. There are multiple influences on the equity of proximity to healthcare services. The issues identified can be largely explained by a lack of healthcare services in the urban peripheries and by high disequilibrium in the PT network in the inner city. This reflects the suburbanization of poverty that many cities around the world have been experiencing.

**Author Contributions:** Conceptualization, Maohua Liu and Xishihui Du; methodology, Xishihui Du; software, Maohua Liu; resources, Maohua Liu; writing—original draft preparation, Siqi Luo; writing—review and editing, Xishihui Du and Siqi Luo; visualization, Siqi Luo; supervision, Maohua Liu. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was supported by the Social Science Planning Fund of Liaoning Province under Grant number L19CSH001.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** The data presented in this study are available on request from the corresponding author.

**Acknowledgments:** The authors acknowledge the contribution of all the anonymous reviewers that improved the quality of the paper.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**

