Spatial Accessibility Patterns to Public Hospitals in Shanghai: An Improved Gravity Model

Abstract

The spatial distribution of public hospitals is a crucial factor in patient access and thus important to health equity, especially amid the COVID-19 pandemic crisis. In China, a major issue is to improve accessibility to public health care in the face of the limited space for further growth of cities. Therefore, this study assesses the current spatial accessibility patterns to public hospitals in Shanghai, using point density analysis and the comprehensively improved gravity model. The results identify the challenges of inadequate and imbalanced spatial distribution of public hospitals at the district level and show that the central urban area has more high-quality public health care resources and also has higher accessibility levels. Based on the comparisons in each district, the study also summarizes four policy implications for future public hospital and public health care system optimization in Shanghai, which can also be leveraged for the planning of public hospitals in other cities or other public service facilities. The first is to deploy public hospitals in areas with a large population and to introduce branches of top-level hospitals to noncentral areas. The second is the full utilization of community health service centers to relieve the burden of public hospitals to improve accessibility.

公立医院的空间分布是患者得到治疗的重要因素, 对健康公平至关重要—尤其是在新冠病毒流行期间。在中国, 主要的问题是：在有限的城市发展空间里去改善公共健康服务的可及性。本研究采用点密度分析和全面改进的重力模型, 评估了上海市公立医院的空间可及性模式。研究结果指出, 区级公立医院的空间分布面临着不充分和不平衡的挑战, 城市中心有更多高质量的公共健康服务资源、更高的可及性。基于对上海市各区的比较, 本研究还总结出了优化上海市未来公立医院和公共健康服务体系的四个政策, 也可用于其它城市公立医院或其它公共服务设施的规划。一是在人口众多的地区设立公立医院, 在非中心地区引入高等级医院的分支机构。二是充分利用社区健康服务中心, 减轻公立医院的负担, 提高可及性。

La distribución espacial de los hospitales públicos es un factor crucial para el acceso de los pacientes y, por lo mismo, importante para la equidad sanitaria, en especial en medio de la crisis de la pandemia del COVID-19. En China, una de las principales cuestiones es mejorar la accesibilidad a la salubridad pública frente a las restricciones de espacio para el mayor crecimiento de las ciudades. Por tanto, este estudio evalúa los actuales patrones de acceso espacial a los hospitales públicos en Shanghái, usando el análisis de densidad de puntos y el modelo de gravedad, notable e íntegramente mejorado. Los resultados identifican los desafíos que presenta una distribución espacial inadecuada y desequilibrada de los hospitales públicos, a nivel distrital, y muestran que el área urbana central cuenta con mayores recursos de servicios de salud pública de alta calidad y tiene también más altos niveles de accesibilidad.

Key Words:

• accessibility analysis
• China
• improved gravity model
• public hospital
• spatial distribution

关键词:

• 可及性分析
• 中国
• 改进的重力模型
• 公立医院
• 空间分布。

Palabras clave:

• análisis de accesibilidad
• China
• distribución espacial
• hospital público
• modelo de gravedad mejorado

Access to health services is a fundamental need that matters for all people (World Health Organization 2017). Different from other needs such as food and shelter, health care has a distinct impact on opportunity equality as disease and disability can restrict the range of opportunities open to individuals (Daniels 1983; C. Yin et al. 2018). Moreover, health care is usually scarcer in quantity, more unequal in distribution, and more expensive than other needs (Daniels 2001; Pan et al. 2016). Therefore, the importance and scarcity of health care jointly render the equity in health care distribution, mainly realized by hospitals, a critical social justice issue. In addition to health care resource allocation inside hospitals, the spatial distribution of hospitals, a crucial factor in patient access, has been widely recognized as important to determine not only health outcomes but also health equity (Braveman 2006; Buchmueller, Jacobson, and Wold 2006; World Health Organization 2020). Especially amid the recent COVID-19 pandemic, there is an urgent need to provide spatially balanced and accessible health facilities in urban areas (Yao et al. 2021). Thus, researchers worldwide, including but not limited to western Asia, Europe, and the United States, have put particular focus on the spatial accessibility patterns of health care facilities, and many of them identified that there could be a considerable variation in terms of the spatial distribution and accessibility of health care facilities, regardless of the study methods they used (Fransen et al. 2015; Mansour 2016; P. Yin 2019).

Relevant studies on the spatial distribution of hospitals mainly focus on four aspects: (1) comparison of accessibility level to hospitals among different regions of a certain population residence (Hou and Jiang 2014; G. Cheng et al. 2016; Reshadat et al. 2019), (2) comparison of accessibility level to different kinds of hospitals among a certain population (Alam et al. 2013; Pan et al. 2016), (3) the spatial distribution analysis of hospitals among a certain population to treat a certain type of disease (De Mello-Sampayo 2016; L. Liu and Wang 2020), and (4) comparison of accessibility level to hospitals by different travel modes (S. Zheng et al. 2019; Y. Zhang et al. 2020). For example, G. Cheng et al. (2016) explored the difference in spatial accessibility to high-level hospitals with a kernel density two-step floating catchment area method and found that accessibility in south and central parts of Shenzhen, a megacity in southern China, was significantly higher than that in the eastern and western parts (G. Cheng et al. 2016). Pan et al. (2016) combined the results of the nearest-neighbor method and the enhanced two-step floating catchment area (E2SFCA) method and found that public hospitals surpassed private hospitals in spatial accessibility, whereas the west Sichuan Province of China only relied on public health care (Pan et al. 2016). L. Liu and Wang (2020) explored the spatial distribution of 117 fever clinics in Shanghai against the background of COVID-19 and found that the density of fever clinics in suburban areas is markedly lower than that in the central urban area (L. Liu and Wang 2020). Zheng et al. (2019) discovered that there are significant differences in hospital accessibility and travel time using public transit and self-driving modes of transportation (Zheng et al. 2019).

Existing studies frequently use accessibility level indicators to quantitatively measure the rationality of spatial distribution (Kanuganti, Sarkar, and Singh 2016; Tan and Guan 2021). Accessibility can be defined as the difficulty of reaching certain facilities from somewhere based on a given way of transportation, in which the difficulty is decided by a multidimensional set of factors including the characteristics of providers, the preferences, and the affordability of consumers’ transportation networks (Dalvi and Martin 1976; Jiao et al. 2012; Guan 2019). The most commonly used methods in calculating the accessibility level are E2SFCA models (W. Luo and Qi 2009; Ngui and Vanasse 2012), network analysis models (Tao et al. 2018), and improved gravity models (Yiannakoulias, Bland, and Svenson 2013). In these methods, travel time is always one of the variables factored into the accessibility level. Although travel time is usually estimated in geographic information systems (GIS) by setting the travel cost for different kinds of roads (Yang et al. 2016), some recent studies have indicated the advantages of using travel time generated by Web mapping application programming interfaces (APIs) to measure travel impedance (Socharoentum and Karimi 2015; S. Zhang et al. 2019). Taking various factors such as speed limits, waiting times, and traffic jams into consideration, the results can be more accurate than using a fixed, hypothetical impedance coefficient to predict travel time and can also better simulate route planning in real life than using road network analysis (Zheng et al. 2019).

In China, along with the rapid economic development in the past decades, health care provision has also improved but not enough to cope with large urbanized populations with relatively limited medical resources (S. Liu and Griffiths 2011). Public hospitals aim to provide basic health care to cover the entire population (National Health Commission of the People’s Republic of China [NHCPRC] 2018, 2020), and thus the current national health care insurance covers partial treatment fees incurred in public hospitals to reduce the costs for public health care (Shanghai Municipal People’s Government [SMPG] 2019). Moreover, the average number of patient beds of public hospitals in urban areas of large cities is often below national averages due to the higher density of population served, which is contrary to common sense, as shown in Appendix B, Table B.1 (NHCPRC 2019). Consequently, the spatial distribution of public hospitals, an important component of the national health system, is a critical issue of public policymaking (Zhu et al. 2019).

There are two main research gaps in the existing studies. The first is that although integrating big data–based Web mapping APIs into accessibility measurement is still in its burgeoning stages, most of the studies that use the Web mapping API directly use the travel time generated from a Web mapping API as the accessibility measurement or put it into simplified measurement models that neglect many other factors such as the class, capacity of the hospital, and the population (Guan et al. 2021). The second is that many of the existing studies conduct either city-level or district-level analysis on the spatial distribution and accessibility of health care facilities and stop at identifying the localized pattern for one particular city or region. The study design renders the results not deep enough because, without sensitivity analysis, it cannot dig out the fundamental problem and the targeted solution. Also, given that the fundamental problem and relationship have not been revealed, the results generated are localized, and thus the policy implications offered cannot be widely leveraged to other places.

In view of these facts, this study examines both the spatial distribution and the accessibility pattern of public hospitals in Shanghai for a comprehensive analysis. The study first uses point density analysis to directly examine the spatial distribution characteristics of public hospitals in Shanghai. More important, this study further examines the rationale of the public hospital spatial distribution from the patient perspective through the accessibility indicator calculated with the improved gravity model and the travel time generated by the Web mapping API. Compared with existing studies, this study closes the research gap by making contributions from the following two perspectives. First, this study integrates the big data–based Web mapping API into an improved gravity model that considers the class, capacity, and nearby population’s competition of public hospitals to improve the precision of accessibility analysis without sacrificing comprehensiveness. Such methodology has seldom been seen in relevant studies especially in the Chinese context. Second, this study follows a sensitivity analysis approach in the accessibility analysis from city level to district level. Many of the previous studies end with the analysis of the existing spatial distribution patterns (Ngui and Vanasse 2012; Zheng et al. 2019). This study summarizes the challenges and experiences in the case of Shanghai based on accurate and in-depth identification. Furthermore, this study offers suggestions to better the spatial distribution of public hospitals in Shanghai. The framework of this study could be leveraged for spatial planning of public hospitals in other cities or other public service facilities due to their commonality.

Materials and Methods

Study Area and Data

The study area of Shanghai, shown in Figure 1,¹ was selected using the 2019 Shanghai Administrative Map in the 2019 Shanghai Almanac (Office of Shanghai Chronicles [OSC] 2020). The study area covers all sixteen districts of Shanghai, including seven districts in the central urban area (i.e., Huangpu, Xuhui, Changning, Jing’an, Putuo, Hongkou, and Yangpu), one district in the central-suburban area (i.e., Pudong), and eight districts in the suburban area (i.e., Minhang, Baoshan, Jiading, Jinshan, Fengxian, Songjiang, Qingpu, and Chongming). The central-suburban area and the suburban area are collectively referred to as the noncentral area. There are 214 subdistricts in total, the minimum research unit of this study, including 107 streets or Jiedao, 105 towns or Zhen, and two townships or Xiang (Office of Shanghai Chronicles 2020). The subdistrict government office is defined as the location of each population point, where all residents in the subdistrict are supposed to depart from. The population of the subdistricts was obtained from the 2019 statistical yearbook of each district. The travel time used in this study is obtained through Baidu’s Web mapping API (see https://map.baidu.com/).²

Figure 1 The study area of Shanghai.

Methods for the Point Density Analysis and Comprehensively Improved Gravity Model

This study applied the point density analysis in GIS to visualize the spatial distribution of public hospitals. The point density analysis calculates the density of point features by totaling and dividing the number of points that fall within the area of the neighborhood of a cell. The accessibility indicator for each population point was calculated using the improved gravity model. The research design framework is summarized in a flowchart in Figure 2. The gravity model is expressed in Equation 1(1) $$A_i=\sum _{j=1}^n{A}_{ij}=\sum _{j=1}^n\frac{M_j}{D_{ij}^{\beta }},$$(1) : (1) $$A_i=\sum _{j=1}^n{A}_{ij}=\sum _{j=1}^n\frac{M_j}{D_{ij}^{\beta }},$$(1) where $A_i$ is the aggregated accessibility index of population point i to all reachable hospitals, $A_{ij}$ is the accessibility index of population i to reachable hospitals j, n is the quantity of destination j, $M_j$ is the scale of j, and $D_{ij}^{\beta }$ is the impedance between i and j when the friction coefficient is $\beta .$ Although $\beta$ is necessary when using the network analysis in GIS to quantify the estimated travel friction, it is no longer needed because the travel time generated from the big data–based Web mapping API has already considered the friction.

Figure 2 Flowchart of the research design.

Based on Equation 1(1) $$A_i=\sum _{j=1}^n{A}_{ij}=\sum _{j=1}^n\frac{M_j}{D_{ij}^{\beta }},$$(1) , Weibull (1976) improved the gravity model by integrating the demand competition among potential patients into the model because health care resources are generally limited (Weibull 1976). The improved model is expressed in Equations 2(2) $$A_i=\sum _{j=1}^n{A}_{ij}=\sum _{j=1}^n\frac{M_j}{D_{ij}^{\beta }{V}_j}$$(2) and 3(3) $$V_j=\sum _{i=1}^m\frac{P_i}{D_{ij}^{\gamma }},$$(3) : (2) $$A_i=\sum _{j=1}^n{A}_{ij}=\sum _{j=1}^n\frac{M_j}{D_{ij}^{\beta }{V}_j}$$(2) (3) $$V_j=\sum _{i=1}^m\frac{P_i}{D_{ij}^{\gamma }},$$(3) where $V_j$ is the competition among population points around the reachable scale of hospital j, m is the number of reachable population points around hospital j, $P_i$ is the population quantity of the population point i, and $D_{ij}^{\gamma }$ is the impedance between i and j when the friction coefficient is $\gamma .$ Similar to $\beta ,$ $\gamma$ is no longer needed because the travel time generated from the big data–based Web mapping API has already considered the friction.

Although the scale and the service level of a hospital have a positive correlation with its attractiveness to potential patients, in Chinese hospital systems, they can be reflected through the class level of the hospital. Therefore, this study further improves Equation 2(2) $$A_i=\sum _{j=1}^n{A}_{ij}=\sum _{j=1}^n\frac{M_j}{D_{ij}^{\beta }{V}_j}$$(2) into Equation 4(4) $$A_i=\sum _{j=1}^n{A}_{ij}=\sum _{j=1}^n\frac{{\alpha }_j{S}_j}{D_{ij}^{\beta }{V}_j},$$(4) : (4) $$A_i=\sum _{j=1}^n{A}_{ij}=\sum _{j=1}^n\frac{{\alpha }_j{S}_j}{D_{ij}^{\beta }{V}_j},$$(4) where the $M_j$ in Equation 2(2) $$A_i=\sum _{j=1}^n{A}_{ij}=\sum _{j=1}^n\frac{M_j}{D_{ij}^{\beta }{V}_j}$$(2) is calculated by multiplying ${\alpha }_j$ by $S_j,$ in which ${\alpha }_j$ is the class coefficient of hospital j and $S_j$ represents its service ability.

To be more accurate and aligned with the practical traveling situation, this study further improves the gravity model by using the $T_{ij}$ generated by the Baidu Web mapping API to measure the travel time from population point i to public hospital j (or reverse) given all impedance factors instead of the traditional network analysis module in GIS.

The final version of the improved gravity model used in this study is expressed in Equations 5(5) $$A_i=\sum _{j=1}^n{A}_{ij}=\sum _{j=1}^n\frac{{\alpha }_j{S}_j}{T_{ij}{V}_j}$$(5) and 6(6) $$V_j=\sum _{i=1}^m\frac{P_i}{T_{ij}}.$$(6) : (5) $$A_i=\sum _{j=1}^n{A}_{ij}=\sum _{j=1}^n\frac{{\alpha }_j{S}_j}{T_{ij}{V}_j}$$(5) (6) $$V_j=\sum _{i=1}^m\frac{P_i}{T_{ij}}.$$(6)

The improved gravity model considers various factors in the process of access: the location, class, and scale of the hospitals; potential competition within and among population points; and traffic conditions. $S_j$ is quantified by the number of hospital patient beds. Considering the actual situation in Shanghai in which there are only five classes available for public health institutions, ${\alpha }_j$ is 5 when public hospital j is a 3A, 4 when public hospital j is a 3B, 3 when public hospital j is a 2A, and 2 when public hospital j is a 2B. $P_i$ is the 2018 year-end population of subdistrict i. This study uses the interrupt travel times for $T_{ij}$ based on the class coefficient of public hospitals adopted from M. Cheng and Lian (2018): no limitation for third-class hospitals and $T_{ij}\le 30$ minutes for second-class hospitals. M. Cheng and Lian (2018) conducted a survey among residents for the interrupted travel time.³

Results

Spatial Distribution of Public Hospitals

Overall, public hospitals in Shanghai are disproportionately distributed, as shown in Figure 3. More specifically, the spatial distribution of public hospitals is dense in the central urban area and sparse in the suburbs. The spatial distribution of community health care centers is shown in Figure 4, which provides general information on overall health service facilities in Shanghai. The density of public hospitals and the occupancy of public hospitals per thousand people by district are given in Table 1.⁴

Figure 3 Density distribution of public hospitals in Shanghai.

Figure 4 The community health service centers in Shanghai.

Table 1 Spatial distribution of public hospitals in Shanghai

District	Property	Population (%)	Acreage (%)	Public hospital				Density of public hospitals (place/km²)	Occupancy of public hospitals (place/thousand people)
				Third class	Second class	Total	%
Huangpu	Central	2.70	0.32	7	7	14	8.43	0.6843	0.0214
Xuhui	Central	4.47	0.86	9	3	12	7.23	0.2191	0.0111
Changning	Central	2.86	0.60	2	5	7	4.22	0.1828	0.0101
Jing’an	Central	4.38	0.58	8	11	19	11.45	0.5152	0.0179
Putuo	Central	5.29	0.86	3	4	7	4.22	0.1277	0.0055
Hongkou	Central	3.29	0.37	4	7	11	6.63	0.4685	0.0138
Yangpu	Central	5.42	0.96	6	6	12	7.23	0.1976	0.0091
Pudong	Central-suburban	22.90	19.09	16	10	26	15.66	0.0215	0.0047
Minhang	Suburban	10.49	5.85	7	5	12	7.23	0.0324	0.0047
Baoshan	Suburban	8.43	4.27	3	7	10	6.02	0.0369	0.0049
Jiading	Suburban	6.56	7.32	2	5	7	4.22	0.0151	0.0044
Jinshan	Suburban	3.32	9.24	2	4	6	3.61	0.0102	0.0075
Fengxian	Suburban	4.75	10.84	1	4	5	3.01	0.0073	0.0043
Songjiang	Suburban	7.27	9.55	2	6	8	4.82	0.0132	0.0045
Qingpu	Suburban	5.03	10.57	1	3	4	2.41	0.0060	0.0033
Chongming	Suburban	2.84	18.70	1	5	6	3.61	0.0051	0.0087
Shanghai		100.00	100.00	74	92	166	100.00	0.0262	0.0068
	Central	28.41	4.56	39	43	82	49.40	0.2833	0.0119
	Central-suburban	22.90	19.09	16	10	26	15.66	0.0215	0.0049
	Suburban	48.69	76.34	19	39	58	34.94	0.0120	0.0049

Taking acreage and population into consideration, the imbalance of the spatial distribution of public hospitals in Shanghai becomes more significant (see Table 1). With merely 4.56 percent of the total acreage and 28.41 percent of the total population, the central urban area contains eighty-two public hospitals, nearly half the total number. Although the central districts generally have a higher population density, based on the indicator of public hospitals per thousand people, the central urban area still has a higher share of public hospitals. The occupancy of public hospitals normalized by population in the central urban area is 1.75 times higher than the Shanghai average and roughly 2.5 times higher than the average level of the noncentral area. In addition to the imbalance in quantity, the quality of public hospitals in the central urban area is generally higher than that in the noncentral area. Furthermore, among the seventy-four third-class public hospitals in Shanghai, thirty-nine are in the central urban area, accounting for more than half of the total. Therefore, in general, the spatial distribution of public hospitals in Shanghai is imbalanced, with the central urban area housing abundant and high-quality public hospitals and the noncentral area, with larger acreage, more population, and yet lower density and lower occupancy per capita of public hospitals, containing public hospitals that are insufficient in both quantity and quality.

Accessibility Analysis

City-Level Accessibility Analysis. The city-level accessibility analysis is based on the scenario that patients choose and go to a public hospital after considering the class, scale, and travel time. Therefore, this part of the analysis is to examine the balance and rationality of spatial distribution of public hospitals from the perspective of patients, the actual users of public hospitals, through the indicator of accessibility to public hospitals. In combination with the definition of travel time $T_{ij},$ this part of the analysis further requires that, in Equation 5(5) $$A_i=\sum _{j=1}^n{A}_{ij}=\sum _{j=1}^n\frac{{\alpha }_j{S}_j}{T_{ij}{V}_j}$$(5) , all third- and second-class public hospitals within a thirty-minute drive from the population point in Shanghai are regarded as accessible; in Equation 6(6) $$V_j=\sum _{i=1}^m\frac{P_i}{T_{ij}}.$$(6) , all population points are considered as accessible for the third-class public hospital, and all population points that are within a thirty-minute drive from the second-class public hospital are considered accessible for the second-class public hospital.

As shown in Figure 5, the accessibility level to public hospitals in Shanghai varies from high (>315) to low (<125) from the central urban area to the noncentral area. The areas with the lowest accessibility level are not necessarily in the periphery of the suburban area, however. Besides northern Chongming, which has a large area with the lowest accessibility level, other lowest level areas are scattered in the noncentral area. In addition, there are high points of accessibility in the noncentral area, some of which are at the border of districts. In addition, the disparity in accessibility level is higher in the central urban area than in the noncentral area. Taking the spatial distribution of public hospitals into consideration, areas with dense location of public hospitals tend to have higher accessibility levels. Therefore, compared with residents in the noncentral area, central urban residents have quantitatively more accessible public health care resources. The competition indicators are shown in Table B.2.

Figure 5 The city-level accessibility to public hospitals in Shanghai.

Table B.2 Statistical summary of public hospitals in Shanghai

Public hospital	Patient beds	Class	Class coefficient	City-level competition indicator	District-level competition indicator
Shanghai Changzheng Hospital	980	3A	5	0.801	0.108
The South Branch of Shanghai No. 9 People’s Hospital	1,000	3A	5	0.829	0.184
The West Branch of Shanghai Renji Hospital	460	3A	5	0.873	0.141
Shanghai Ruijin Hospital	1,800	3A	5	0.780	0.110
The West Branch of Shanghai Shuguang Hospital	600	3A	5	0.815	0.158
The East Branch of Shanghai Oral Disease Prevention and Control Hospital	17	3A	5	1.071	0.323
The Huangpu Branch of Shanghai No. 9 People’s Hospital	700	2A	3	0.466	0.146
The Luwan Branch of Shanghai Ruijin Hospital	426	2A	3	0.674	0.162
Shanghai Xiangshan Traditional Chinese Medicine Hospital	102	2A	3	0.589	0.137
Shanghai Huangpu District Traditional Chinese Medicine Hospital	404	2A	3	0.509	0.153
The Huangpu Branch of Obstetrics & Gynecology Hospital of Fudan University	820	3A	5	0.822	0.150
Shanghai Huangpu District Infectious Disease Hospital	140	2A	3	0.567	0.160
Shanghai Huangpu District Mental Health Hospital	600	2A	3	0.699	0.124
Shanghai Ruijin Rehabilitation Hospital	100	2B	2	0.460	0.122
Shanghai Zhongshan Hospital	2,005	3A	5	0.760	0.122
Shanghai No. 6 People’s Hospital	1,700	3A	5	0.647	0.100
Shanghai Longhua Hospital	550	3A	5	0.708	0.137
Shanghai Eye & Ent Hospital	374	3A	5	0.718	0.085
Shanghai Cancer Hospital	800	3A	5	0.705	0.114
Shanghai Chest Hospital	580	3A	5	0.701	0.088
The West Branch of Shanghai Oral Disease Prevention and Control Hospital	17	3A	5	0.734	0.082
Shanghai No. 8 People’s Hospital	618	2A	3	0.382	0.134
Shanghai Dahua Hospital	210	2B	2	0.302	0.174
Shanghai Xuhui District Central Hospital	588	3B	4	0.731	0.076
The Xuhui Branch of Shanghai Mental Health Hospital	1,878	3A	5	0.652	0.108
Shanghai Xuhui District Mental Health Hospital	290	2A	3	0.311	0.107
Shanghai Guanghua Integrated Traditional Chinese and Western Medicine Hospital	400	3A	5	0.775	0.111
Shanghai Tongren Hospital	1,200	3B	4	0.693	0.116
Shanghai Electric Power Hospital	307	2A	3	0.454	0.075
Shanghai Civil Aviation Hospital	120	2B	2	0.376	0.091
No. 905 Hospital of the Chinese People’s Liberation Army Navy	550	2A	3	0.485	0.077
Shanghai Tianshan Traditional Chinese Medicine Hospital	236	2A	3	0.273	0.093
Shanghai Changning District Mental Health Hospital	320	2A	3	0.191	0.059
Shanghai Huashan Hospital	1,216	3A	5	0.772	0.091
Shanghai Huadong Hospital	1,050	3A	5	0.866	0.114
Shanghai No. 10 People’s Hospital	1,065	3A	5	0.699	0.115
Shanghai Traditional Chinese Medicine Hospital	300	3A	5	0.751	0.135
Shanghai Eye Disease Prevention and Control Hospital	20	3A	5	0.761	0.138
Shanghai Oral Disease Hospital of Tongji University	50	3A	5	0.685	0.099
Shanghai Gonghui Hospital	207	2B	2	0.373	0.095
Shanghai Shibei Hospital	480	2A	3	0.248	0.114
Shanghai Jing’an District Zhabei Central Hospital	669	2A	3	0.499	0.140
Shanghai Jing’an District Central Hospital	645	3B	4	0.707	0.122
Shanghai Post & Telecommunications Hospital	266	2A	3	0.393	0.090
Shanghai Beizhan Hospital	378	2A	3	0.543	0.170
Shanghai Jing’an District Traditional Chinese Medicine Hospital	102	2A	3	0.463	0.147
Shanghai Occupational Disease for Chemical Industry Prevention and Control Hospital	50	2A	3	0.609	0.111
The East Branch of Shanghai Children Hospital	150	3A	5	0.616	0.090
Shanghai No. 3 Rehabilitation Hospital	100	2B	2	0.294	0.083
Shanghai No. 4 Rehabilitation Hospital	300	2B	2	0.295	0.126
Shanghai Jing’an District Mental Health Hospital	216	2A	3	0.237	0.112
Shanghai No. 3 Civil Mental Health Hospital	400	2A	3	0.260	0.089
Shanghai Tongji Hospital	1,100	3A	5	0.723	0.149
Shanghai Putuo District Central Hospital	1,050	3B	4	0.610	0.109
Shanghai Liqun Hospital	500	2A	3	0.190	0.090
Shanghai Putuo District People’s Hospital	784	2A	3	0.399	0.091
Shanghai Putuo District Traditional Chinese Medicine Hospital	184	2A	3	0.208	0.117
Shanghai Children Hospital	550	3A	5	0.592	0.064
Shanghai Putuo District Mental Health Hospital	200	2A	3	0.304	0.139
The North Branch of Shanghai No. 1 Hospital	820	3A	5	0.740	0.091
Shanghai Integrated Traditional Chinese and Western Medicine Hospital	446	3A	5	0.796	0.104
Shanghai Yueyang Traditional Chinese and Western Medicine Hospital	900	3A	5	0.640	0.078
Shanghai No. 4 People’s Hospital	635	2A	3	0.385	0.104
Shanghai Jiangwan Hospital	350	2A	3	0.351	0.101
The Hongkou Branch of Shanghai Changhai Hospital	450	2A	3	0.376	0.131
Shanghai Construction Group Hospital	450	2A	3	0.502	0.089
Shanghai Seaman’s Hospital	450	2A	3	0.520	0.125
Shanghai Waterway Hospital	150	2B	2	0.522	0.104
The Hongkou Branch of Shanghai Public Health Hospital	60	3A	5	0.725	0.090
Shanghai Hongkou District Mental Health Hospital	100	2A	3	0.364	0.084
Shanghai Changhai Hospital	1,800	3A	5	0.758	0.210
Shanghai Xinhua Hospital	1,773	3A	5	0.672	0.128
Shanghai Yangpu District Central Hospital	956	3B	4	0.718	0.190
Shanghai Eastern Hepatobiliary Surgery Hospital	1,000	3A	5	0.742	0.195
Shanghai Lung Hospital	838	3A	5	0.600	0.083
Shanghai Kongjiang Hospital	140	2B	2	0.416	0.213
Shanghai Shidong Hospital	686	2A	3	0.389	0.255
Shanghai Yangpu District Traditional Chinese Medicine Hospital	265	2A	3	0.278	0.102
The Yangpu Branch of Obstetrics & Gynecology Hospital of Fudan University	820	3A	5	0.602	0.101
Shanghai No. 1 Rehabilitation Hospital	500	2B	2	0.286	0.124
Shanghai Yangpu District Mental Health Hospital	300	2A	3	0.308	0.119
The Infectious Disease Branch of Shanghai Yangpu District Central Hospital	100	2A	3	0.312	0.151
The South Branch of Shanghai Renji Hospital	600	3A	5	0.538	0.074
Shanghai No. 5 People’s Hospital	630	3B	4	0.482	0.099
The West Branch of Shanghai Huashan Hospital	800	3A	5	0.566	0.061
The West Branch of Shanghai Eye & Ent Hospital	330	3A	5	0.493	0.064
Shanghai Minhang District Central Hospital	800	3B	4	0.604	0.093
Shanghai Minhang District Integrated Traditional Chinese and Western Medicine Hospital	315	2B	2	0.083	0.056
Shanghai Minhang District Traditional Chinese Medicine Hospital	300	2A	3	0.394	0.108
Shanghai Minhang District Cancer Hospital	120	2A	3	0.078	0.072
Shanghai Children’s Hospital of Fudan University	600	3A	5	0.685	0.096
The Minhang Branch of Shanghai Mental Health Hospital	880	3A	5	0.588	0.128
Shanghai No. 1 Civil Mental Health Hospital	230	2B	2	0.207	0.070
Shanghai Minhang District Mental Health Hospital	400	2B	2	0.040	0.015
The North Branch of Shanghai Huashan Hospital	600	3A	5	0.639	0.103
The North Branch of Shanghai No. 9 People’s Hospital	800	3A	5	0.638	0.096
Shanghai Baoshan District Integrated Traditional Chinese and Western Medicine Hospital	504	3A	5	0.562	0.083
The Baoshan Branch of Shanghai No. 1 Hospital	658	2A	3	0.267	0.132
Shanghai Zhongye Hospital	751	2A	3	0.070	0.062
Shanghai Renhe Hospital	360	2A	3	0.198	0.117
Shanghai Dachang Hospital	455	2B	2	0.219	0.071
Shanghai Luodian Hospital	400	2B	2	0.039	0.020
Shanghai Baoshan District Mental Health Hospital	250	2A	3	0.109	0.065
Shanghai No. 2 Rehabilitation Hospital	300	2B	2	0.299	0.109
The North Branch of Shanghai Ruijin Hospital	600	3A	5	0.463	0.049
The Anting Branch of Shanghai Eastern Hepatobiliary Surgery Hospital	1,500	3A	5	0.438	0.050
Shanghai Anting Hospital	200	2B	2	0.156	0.142
Shanghai Nanxiang Hospital	325	2A	3	0.164	0.068
Shanghai Jiading District Central Hospital	800	2A	3	0.046	0.046
Shanghai Jiading District Traditional Chinese Medicine Hospital	204	2A	3	0.037	0.037
Shanghai Jiading District Mental Health Hospital	360	2A	3	0.041	0.032
The East Branch of Shanghai No. 6 People’s Hospital	600	3A	5	0.331	0.097
The South Branch of Shanghai Eastern Hospital	1,000	3A	5	0.720	0.213
Shanghai Eastern Hospital	800	3A	5	0.828	0.198
The Pudong Branch of Shanghai No. 9 People’s Hospital	20	3A	5	0.769	0.250
The East Branch of Shanghai Renji Hospital	1,000	3A	5	0.783	0.264
The North Branch of Shanghai Renji Hospital	190	3A	5	0.767	0.255
The East Branch of Shanghai Huashan Hospital	200	3A	5	0.754	0.237
The East Branch of Shanghai Shuguang Hospital	720	3A	5	0.751	0.236
Shanghai No. 7 People’s Hospital	730	3A	5	0.633	0.184
The Pudong Branch of Shanghai Cancer Hospital	240	3A	5	0.611	0.181
Shanghai Gongli Hospital	600	3B	4	0.807	0.282
The Pudong Branch of Shanghai Longhua Hospital	700	3A	5	0.822	0.241
Shanghai Punan Hospital	550	2A	3	0.495	0.206
Shanghai Pudong New District People’s Hospital	800	3B	4	0.598	0.206
Shanghai Zhoupu Hospital	1,000	3B	4	0.580	0.170
Shanghai Pudong Hospital	1,000	3B	4	0.449	0.145
Shanghai Guangming Traditional Chinese Medicine Hospital	300	2A	3	0.066	0.066
Shanghai Pudong New District Traditional Chinese Medicine Hospital	202	2A	3	0.144	0.141
Shanghai Pudong New District Lung Hospital	100	2A	3	0.095	0.088
Shanghai Children Medical Center	500	3A	5	0.799	0.282
Shanghai Pudong New District Infectious Disease Hospital	100	2A	3	0.075	0.075
Shanghai Pudong New District Mental Health Hospital	450	2A	3	0.310	0.112
Shanghai Nanhua Hospital	116	2A	3	0.081	0.081
Shanghai Pudong New District Nanhui Mental Health Hospital	180	2A	3	0.055	0.055
Shanghai General Prison Hospital	120	2B	2	0.077	0.056
Shanghai Pudong New District Geriatric Hospital	300	2B	2	0.050	0.046
Shanghai Jinshan Hospital	700	3B	4	0.357	0.050
The Jinshan Branch of Shanghai No. 6 Hospital	620	2A	3	0.052	0.048
Shanghai Tinglin Hospital	200	2B	2	0.160	0.144
Shanghai Jinshan District Integrated Traditional Chinese and Western Medicine Hospital	300	2A	3	0.036	0.026
Shanghai Public Health Hospital	600	3A	5	0.381	0.044
Shanghai Jinshan District Mental Health Hospital	300	2A	3	0.039	0.037
Shanghai Fengxian District Central Hospital	1,000	3B	4	0.426	0.040
Shanghai Fengcheng Hospital	400	2A	3	0.090	0.058
Shanghai Fengxian District Traditional Chinese Hospital	400	2A	3	0.044	0.042
Shanghai Guhua Hospital	200	2A	3	0.036	0.034
Shanghai Fengxian District Mental Health Hospital	400	2A	3	0.038	0.029
The South Branch of Shanghai No. 1 People’s Hospital	1,000	3A	5	0.465	0.080
Shanghai Jiuting Hospital	100	2B	2	0.091	0.040
Shanghai Sijing Hospital	300	2B	2	0.041	0.031
Shanghai Songjiang District Central Hospital	805	3B	4	0.387	0.059
Shanghai Fangta Traditional Chinese Medicine Hospital	170	2A	3	0.054	0.051
Shanghai No. 5 Rehabilitation Hospital	150	2B	2	0.112	0.099
Shanghai Songjiang District Mental Health Hospital	200	2A	3	0.053	0.037
Shanghai Yangzhi Rehabilitation Hospital	360	2B	2	0.043	0.041
The Qingpu Branch of Shanghai Zhongshan Hospital	1,000	3B	4	0.386	0.056
Shanghai Zhujiajiao People’s Hospital	200	2B	2	0.028	0.028
Shanghai Qingpu District Traditional Chinese Medicine Hospital	300	2A	3	0.041	0.062
Shanghai Qingpu District Mental Health Hospital	350	2A	3	0.030	0.028
The Chongming Branch of Shanghai Xinhua Hospital	790	3B	4	0.187	0.052
The Chongming Branch of Shanghai No. 3 People’s Hospital	260	2B	2	0.083	0.083
The Chongming Branch of Shanghai No. 10 People’s Hospital	500	2B	2	0.060	0.060
Shanghai Chongming District Infectious Disease Hospital	149	2A	3	0.047	0.047
Shanghai Chongming District Mental Health Hospital	250	2A	3	0.043	0.044
Shanghai Kangle Hospital	30	2B	2	0.039	0.039

District-Level Accessibility Analysis. The district-level accessibility analysis assumes that for patients, only public hospitals in their own district are accessible. In particular, the central urban districts are regarded as an entire district because they have relatively small acreage and vague borders in daily life. This analysis is designed for two reasons. First, although patients will go to public hospitals beyond the district border in reality, a balanced spatial distribution of public hospitals within the district ensures that patients can quickly access public health care resources. Second, district-level accessibility analysis enables in-depth analyses of spatial distribution of public hospitals for each district (Figure 6) and a comparison among districts (Table 2), which is meaningful for making health care–related policies at the district level. In addition, this study further explores experiences and exposes challenges at the district level, to provide suggestions on the future spatial planning of public hospitals. Table B.3 provides a summary of patient bed, hospital class coefficient, and district-level competition indicators.

Figure 6 The district-level accessibility to public hospitals of Shanghai, overall version (merging central district as one).

Table 2 The district-level accessibility indicator of Shanghai

District	Maximum	Minimum	M	SD	Accessibility distribution (acreage %)
					350–637.6	250–350	150–250	75–150	9.7–75
Huangpu	417.1	306.9	357.6	36.3	61.61	38.39	0.00	0.00	0.00
Xuhui	637.6	261.4	436.5	137.1	61.79	38.21	0.00	0.00	0.00
Changning	234.1	110.2	171.5	32.3	0.00	0.00	91.08	8.92	0.00
Jing’an	433.9	230.2	299.3	51.8	12.94	81.38	5.68	0.00	0.00
Putuo	290.2	132.6	180.5	48.0	0.00	18.79	47.28	33.93	0.00
Hongkou	413.8	170.8	267.3	82.7	24.54	31.28	44.18	0.00	0.00
Yangpu	496.5	237.6	360.7	65.7	54.35	30.73	14.92	0.00	0.00
Pudong	250.7	82.9	138.0	40.1	0.00	0.27	33.14	66.59	0.00
Minhang	378.7	163.5	236.8	53.7	8.28	43.58	48.14	0.00	0.00
Baoshan	223.0	66.5	131.8	40.9	0.00	0.00	26.60	57.13	16.27
Jiading	359.1	74.9	233.6	95.2	0.89	41.45	38.83	10.29	8.54
Jinshan	238.0	137.7	179.0	34.1	0.00	0.00	71.43	28.57	0.00
Fengxian	246.3	78.8	148.3	50.5	0.00	0.00	48.20	51.80	0.00
Songjiang	381.2	78.6	147.4	96.7	2.46	11.88	7.10	78.56	0.00
Qingpu	248.8	33.5	98.3	70.0	0.00	0.00	7.78	52.14	40.08
Chongming	293.0	9.7	63.5	64.0	0.00	4.55	0.00	23.87	71.58

Note: The acreage of subdistricts is weighted in the average of accessibility and the percentage of accessibility distribution.

Table B.3 Statistical summary of subdistricts in Shanghai

Subdistrict	Acreage (km²)	Population	City-level accessibility indicator	District-level accessibility indicator
East Nanjing Road Street	2.41	97,000	403.5	358.7
Waitan Street	2.18	94,930	306.9	292.7
Bansongyuan Street	2.87	88,370	355.8	297.4
Xiaodongmen Street	2.59	91,936	328.2	349.5
Yuyuan Street	1.19	90,695	366.0	334.8
Laoximen Street	1.21	90,100	413.7	317.3
Wuliqiao Street	3.09	65,835	323.4	308.8
Dapuqiao Street	1.59	62,782	368.9	341.0
Middle Huaihai Road Street	1.41	88,000	417.1	314.9
No. 2 Ruijin Road Street	1.93	70,200	361.6	294.2
Tianping Road Street	2.68	86,204	433.7	302.2
Hunan Road Street	1.72	47,929	428.8	295.5
Xietu Road Street	3.18	63,833	618.8	377.0
Fenglin Road Street	2.69	97,938	615.6	339.2
Changqiao Street	5.87	94,544	313.6	272.1
Tianlin Street	4.16	78,828	634.5	314.4
Hongmei Road Street	4.26	20,735	274.6	244.9
Kangjian New Village Street	4.07	83,421	408.8	299.1
Xujiahui Street	4.07	97,818	637.6	307.9
Lingyun Road Street	3.58	79,515	282.5	246.5
Longhua Street	6.13	53,600	493.5	299.9
Caohejing Street	5.23	77,218	496.7	317.4
Huajing Town	7.27	37,490	261.4	237.5
Huayang Road Street	2.04	67,000	170.2	306.4
Xinhua Road Street	2.20	64,475	234.1	324.6
Tianshan Road Street	1.83	77,900	154.4	242.5
Hongqiao Street	4.08	49,763	153.7	307.6
Beixinjing Street	1.80	38,334	110.2	202.3
Jiangsu Road Street	1.52	50,400	122.3	260.9
Zhoujiaqiao Street	1.95	45,720	176.9	276.1
Xianxia New Village Street	2.30	71,692	173.7	267.1
Chengjiaqiao Street	7.60	20,196	172.3	222.4
Xinjing Town	11.89	95,057	182.7	229.7
Jiangning Road Street	1.84	74,682	312.8	339.4
No.2 Shimen Road	1.09	38,989	318.8	348.8
West Nanjing Road Street	1.62	51,047	433.9	419.1
Jing'an Temple Street	1.57	38,288	403.3	333.7
Caojiadu Street	1.49	72,587	338.5	335.8
West Tianmu Road Street	1.94	32,980	327.0	403.0
Beizhan Street	1.74	65,945	318.0	381.9
Baoshan Road Street	1.62	80,266	313.9	347.7
West Zhijiang Road Street	1.64	70,080	351.0	344.1
Gonghexin Road Street	2.72	72,494	277.4	334.0
Daning Road Street	6.24	63,623	280.7	355.5
Pengpu New Village Street	3.81	121,027	279.4	299.1
Linfen Road Street	2.12	58,342	230.2	281.8
Pengpu Town	7.89	98,907	263.6	304.1
Caoyang New Village Street	2.14	87,328	290.2	309.5
Changfeng New Village Street	5.80	89,968	165.1	308.2
Changshu Road Street	4.02	99,257	180.3	306.7
Ganquan Road Street	2.34	86,677	265.1	275.7
Shiquan Road Street	3.49	88,493	174.1	234.5
Yichuan Road Street	2.22	87,820	187.6	346.5
Wanli Street	3.05	39,118	201.9	284.0
Zhenru Street	5.95	117,783	260.1	326.2
Changzheng Town	7.66	95,476	184.1	251.6
Taopu Town	18.83	102,522	132.6	249.3
Ouyang Road Street	1.67	60,754	281.9	318.0
Quyang Road Street	3.04	84,471	296.3	292.7
Guangzhong Road Street	2.91	88,910	183.3	275.5
Jiaxing Road Street	2.63	98,983	267.3	327.6
Liangcheng New Village Street	3.24	70,384	196.2	297.8
North Sichuan Road Street	1.78	83,099	372.5	342.9
Beiwaitan Street	3.98	152,153	413.8	370.3
Jiangwan Street	4.22	90,731	170.8	295.6
Dinghai Road Street	4.44	86,062	496.5	288.1
Pingliang Road Street	3.53	87,142	313.9	281.7
Jiangpu Road Street	2.39	73,378	324.3	300.0
Siping Road Street	2.75	96,278	300.2	258.9
Kongjiang Road Street	2.15	84,085	342.5	294.8
Changbai New Village Street	3.05	60,013	312.5	266.0
Yanji New Village Street	2.05	72,284	361.1	279.5
Yinhang Street	7.98	136,527	379.7	312.8
Daqiao Street	3.99	118,236	349.8	302.4
Wujiaochang Street	7.60	117,968	373.2	375.5
New Jiangwan Town Street	8.67	29,182	237.6	238.2
Wujiaochang Town	9.52	112,983	451.9	372.5
Jiangchuan Road Street	25.85	133,700	265.5	183.8
Gumei Road Street	6.50	71,447	191.6	210.0
Xinhong Street	23.99	33,356	163.5	161.7
Pujin Street	23.99	40,114	226.4	232.5
Xinzhuang Town	19.12	152,135	217.8	211.3
Qibao Town	19.62	142,695	180.3	177.3
Zhuanqiao Town	32.29	71,798	378.7	210.8
Huacao Town	28.50	49,940	270.9	179.7
Hongqiao Town	11.08	76,899	206.5	228.6
Meilong Town	28.07	131,827	187.2	209.6
Wujing Town	37.15	58,533	251.5	222.6
Maqiao Town	37.54	42,082	173.3	164.4
Pujiang Town	78.51	109,271	255.1	204.1
Youyi Road Street	10.45	94,273	223.0	240.2
Wusong Street	7.52	71,578	211.7	256.1
Zhangmiao Street	5.19	118,183	156.2	283.1
Yanghang Town	39.70	70,001	143.9	271.2
Yuepu Town	44.37	69,291	139.2	164.5
Luodian Town	44.19	66,455	133.2	200.0
Luojing Town	48.00	30,223	66.5	104.1
Gucun Town	41.66	113,267	157.7	220.6
Dachang Town	27.22	172,286	105.0	314.7
Miaohang Town	5.96	41,745	109.9	223.9
Songnan Town	13.65	68,429	152.7	223.9
Gaojing Town	7.10	71,193	143.4	296.3
Xincheng Road Street	5.14	27,733	247.2	273.4
Zhenxin Street	5.25	49,334	100.1	200.1
Jiading Town Street	4.10	57,811	359.1	308.6
Nanxiang Town	33.38	68,333	284.0	326.3
Anting Town	89.02	98,271	349.3	202.9
Malu Town	57.09	69,238	232.5	248.8
Xuhang Town	39.95	31,267	159.8	204.2
Huahang Town	39.55	24,546	74.9	111.2
Waigang Town	50.92	31,408	343.3	233.3
Jiangqiao Town	42.40	84,105	114.7	180.9
Weifang New Village Street	3.74	94,018	250.7	247.4
Lujiazui Street	6.90	116,601	226.4	249.9
Zhoujiadu Street	5.52	108,634	217.8	285.0
Tangqiao Street	3.86	57,825	211.6	275.4
Shanggang New Village Street	7.56	93,681	210.4	338.3
Nanmatou Road Street	4.25	78,001	201.6	271.9
Hudong New Village Street	5.39	73,010	196.5	198.3
Jinyang New Village Street	8.02	133,420	193.6	227.0
Yangjing Street	7.38	112,973	187.2	228.8
Puxing Road Street	6.25	106,876	180.4	192.0
Dongming Road Street	4.90	71,683	177.7	243.1
Huamu Street	20.93	135,588	174.2	206.1
Nanhui New Town	152.23	33,005	172.5	126.5
Chuansha New Town	93.35	160,293	172.4	174.7
Gaoqiao Town	39.07	91,191	164.0	163.2
Beicai Town	23.71	145,432	162.4	204.9
Heqing Town	43.03	60,266	162.3	143.7
Tang Town	32.32	53,175	154.8	194.4
Caolu Town	46.54	78,335	149.5	176.0
Jinqiao Town	25.51	31,399	148.0	182.9
Gaohang Town	22.85	64,816	140.2	207.3
Gaodong Town	35.76	37,832	137.3	165.8
Zhangjiang Town	45.02	90,487	130.3	210.8
Sanlin Town	34.15	152,633	128.9	254.1
Huinan Town	65.83	122,396	126.0	220.6
Zhoupu Town	43.68	93,802	125.6	181.1
Xinchang Town	53.44	57,427	124.3	198.3
Datuan Town	50.57	66,946	121.7	190.2
Kangqiao Town	41.55	79,548	120.3	211.9
Hangtou Town	59.51	74,846	120.1	186.6
Zhuqiao Town	160.19	140,271	117.2	209.1
Nicheng Town	59.27	57,427	116.6	118.2
Xuanqiao Town	45.81	44,217	115.3	210.6
Shuyuan Town	54.20	51,658	110.8	118.2
Wanxiang Town	23.14	25,155	84.4	155.1
Laogang Town	66.72	34,510	82.9	150.3
Shihua Street	19.13	51,485	139.3	172.4
Lvgang Town	59.47	41,785	161.9	119.8
Langxia Town	46.56	30,977	148.6	153.0
Zhujing Town	77.10	87,103	238.0	274.1
Jinshanwei Town	55.06	50,429	137.7	122.9
Fengjing Town	91.67	63,360	171.9	213.1
Caojing Town	56.92	30,774	146.0	148.1
Zhangyan Town	34.93	28,745	208.9	163.9
Shanyang Town	43.86	53,718	231.0	164.0
Tinglin Town	79.12	57,443	173.2	192.4
Yueyang Street	5.80	67,818	291.4	231.4
Yongfeng Street	24.39	44,579	348.3	195.8
Fangsong Street	14.76	75,042	381.2	229.8
Guangfulin Street	19.05	26,021	224.6	168.6
Zhongshan Street	41.03	38,582	313.4	218.9
Jiuliting Street	6.79	19,072	86.1	163.1
Sijing Town	23.48	42,650	163.1	197.8
Sheshan Town	66.80	39,380	122.8	191.0
Chedun Town	45.30	36,253	99.0	139.8
Xinqiao Town	35.48	34,615	140.4	182.7
Dongjing Town	24.51	13,875	139.5	183.0
Jiuting Town	24.54	36,542	96.9	172.1
Maogang Town	57.28	38,637	110.9	183.7
Shihudang Town	44.18	27,702	128.3	196.0
Xinbang Town	44.74	26,003	88.1	166.7
Yexie Town	72.54	49,484	78.6	150.4
Xiaokunshan Town	48.70	24,196	120.7	180.6
Xiayang Street	35.71	52,493	248.8	273.5
Yingpu Street	16.36	45,751	238.9	290.0
Xianghuaqiao Street	62.18	38,068	75.4	175.7
Zhujiajiao Town	138.00	61,097	128.5	189.6
Liantang Town	93.66	54,826	69.7	160.0
Jinze Town	108.49	62,702	105.7	168.1
Zhaoxiang Town	40.47	28,695	100.0	176.5
Xujing Town	38.55	39,251	43.9	167.8
Huaxin Town	47.60	38,578	33.5	143.8
Zhonggu Town	30.00	20,659	73.2	179.4
Baihe Town	58.57	47,080	52.6	169.6
Nanqiao Town	85.89	79,582	246.3	255.9
Fengcheng Town	109.91	90,055	78.8	121.0
Zhuanghang Town	69.90	46,524	118.2	176.8
Jinhui Town	71.72	56,057	79.5	179.0
Situan Town	72.92	64,648	176.0	205.3
Qingcun Town	73.10	53,383	153.8	182.1
Zhelin Town	99.05	56,947	182.0	195.9
Haiwan Town	111.84	11,281	146.7	172.0
Xidu Street	28.77	38,593	117.4	219.0
Fengpu Street	15.51	19,816	164.5	211.0
Lvhua Town	37.45	8,627	40.5	74.6
Xincun Township	32.50	10,658	29.2	64.1
Sanxing Town	68.17	39,823	66.0	98.6
Miao Town	95.53	56,533	122.7	150.9
Gangxi Town	45.73	27,789	110.3	131.5
Jianshe Town	42.40	31,340	96.8	129.1
Xinhe Town	57.25	45,807	78.1	123.7
Shuxin Town	58.84	40,401	52.3	109.7
Gangyan Town	75.00	51,864	49.6	112.4
Xianghua Town	53.78	31,221	35.7	117.1
Zhongxing Town	55.00	30,305	34.9	125.7
Chenjia Town	94.90	59,809	17.1	103.5
Chengqiao Town	58.00	88,631	293.0	203.5
Bao Town	63.48	59,842	114.4	178.4
Changxing Town	160.60	40,727	14.2	129.3
Hengsha Township	51.74	33,310	9.7	76.8
Dongping Town	119.70	11,202	46.5	95.3
Xinhai Town	105.04	10,742	35.9	70.6

The overall accessibility level of the central districts is higher than the suburban districts. The numbers of public hospitals in Huangpu, Xuhui, and Yangpu are generally sufficient. These are also the districts with the highest accessibility level citywide.⁵ Comparing the city-level analysis and district-level analysis, one noteworthy point is that the low accessibility of Changning and Putuo districts is disguised in the city-level analysis, in which patients in these two districts are assumed to be able to visit all accessible hospitals within the city. This indicates that Changning and Putuo rely heavily on nearby districts’ public hospital resources. This can be proved when all central urban districts are regarded as a whole. As shown in Appendix A, the accessibility levels of Changning and Putuo are high, which is in sharp contrast to their accessibility levels in Figure 6. The overall accessibility level of the districts in the suburban area is low, and the accessibility pattern varies greatly among districts.⁶

Discussion

As revealed by the results, the spatial distribution and accessibility of public hospitals in Shanghai are generally imbalanced, with the central urban area aggregating more and better hospitals and also enjoying better accessibility and the suburban area having both fewer hospitals and less accessibility. This imbalanced pattern of public hospitals can be largely attributed to the uneven development progress of Shanghai. Most of the central urban areas were actually concession areas in the Treaty Port era, and the current public health care system in the central urban area was largely the legacy of that special period (Yan 2013). In contrast, the suburban area of Shanghai did not rapidly develop from villages into suburbs until the 1990s when the Reform period of opening up and urbanization in Shanghai started. As the suburban area gradually developed and gathered more and more migrant workers, the public health care system then started to develop (Ouyang et al. 2017).

The central urban–suburban divide in public hospitals is dynamic rather than static, however, which can be proven by the study results showing that some suburban districts also have high performance in both spatial distribution and accessibility (e.g., Minhang) and some central urban districts have relatively poor performance (e.g., Changning). Some suburban districts have similar geographic location, population, and number of public hospitals, but the study shows that how they allocate the limited resources can result in different spatial distribution and accessibility. Therefore, the challenges and experiences are not the same based simply on the district attributes.

Although Shanghai is already a highly urbanized and developed city with limited space for planning of new public hospitals (Xiao et al. 2019), further optimization based on the spatial distribution of existing public hospitals should be prioritized. Therefore, the following discussion mainly focuses on the situation of each district given that it is the most suitable administrative level at which to take optimizing measures—small enough to analyze and compare to the actual spatial distribution of public hospitals and large enough to provide intervention via comprehensive planning.

Inadequate Public Hospitals

The shortage of public hospitals is directly reflected in the low density and low occupancy of public hospitals in some districts. For example, Qingpu and Fengxian have the lowest density and occupancy of public hospitals in Shanghai, indicating that existing public hospitals are not enough to meet the needs of patients for public health care services. Due to its small population, Chongming has the highest occupancy level of public hospitals per thousand people, but it still has inadequate public hospitals, as shown by its accessibility indicators. Low population density means it needs more balanced spatial distribution of public hospitals to improve accessibility. These three districts are on the periphery of Shanghai and away from the central urban area, so the insufficiency of public hospitals in these districts directly affects the overall accessibility level of public hospitals. Compared with other major cities in China—for example, Wuhan (0.03 place/km² or below 0.02 place/km² without considering specialized hospitals)—the overall density distribution of hospitals in Shanghai (0.026 place/km²) is similar or slightly better if only considering high-quality hospitals. The inadequate medical resources in suburban districts is a common phenomenon, however. This indicates that this is an issue beyond Shanghai that many other large cities in China are also facing.

Besides the inadequacy of public hospitals, some districts face the problem of the imbalanced class structure of public hospitals, which means lacking public hospitals of a particular class. Although the overall inadequacy is confined to some suburban districts, this problem is more common across Shanghai. On the one hand, third-class public hospitals are inadequate in some districts, which leads to a lack of high-quality health care services. For example, there are only two third-class public hospitals in Changning in the central urban area and only one third-class public hospital in Chongming, Qingpu, and Fengxian in the suburban area. On the other hand, second-class public hospitals are inadequate in some districts. Pudong, for example, has sixteen third-class public hospitals, accounting for roughly 20 percent of the total third-class public hospitals in Shanghai and matching its population and acreage ratio; it only has ten second-class public hospitals, however, accounting for less than 10 percent of the total amount. Although second-class hospitals can also offer basic health care services and target normal diseases, patients are more likely to go to third-class hospitals for higher level health care services because they generally have better medical resources. On the contrary, for more common treatment and services, patients are more likely to go to second-class hospitals because their services offered can service the need and are generally less time-consuming with fewer patients. In this sense, lack of second-class hospitals pushes all patients to have no choice but to turn to third-class hospitals only, leading to longer wait times for regular patients and squeezing the already scarce medical resources for severe patients.

Imbalanced Spatial Distribution of Public Hospitals

The imbalanced spatial distribution of existing public hospitals can be attributed to the rapid urban expansion of Shanghai. As mentioned earlier, more than half of the third-class public hospitals are concentrated in the central urban area, most of which were established before the turn of the century. This leads to a low level and high disparity of accessibility, sometimes even if the public hospitals are generally sufficient. This problem exists in Pudong, Baoshan, and Chongming. For example, the public hospitals in Pudong are mainly concentrated in the western and central areas, which are mainly residential land use with a large population density. In contrast, there is a large area with no public hospital in the southern area, which is mainly under agricultural land use (Shanghai Pudong New District People’s Government 2016). As a result, although the public hospitals are generally sufficient, the overall accessibility level for Pudong, especially its southern area, is still relatively low. Taking into consideration population density, the current locations of hospitals do match the residential land use areas. As urban land in Pudong further expands and new subdistrict centers are being planned in the southern area, such as Nanhui New Town and Shanghai’s Pilot Free Trade Zone in Lingang New Area (Shanghai Pudong New District People’s Government 2016), future planning for new public hospitals should address the current imbalanced distribution of public health care resources. The third-class public hospitals in Baoshan are mainly concentrated in the eastern area along the Yangtze River, which leads to an extremely low accessibility level for public hospitals in the western part of the district, given its general insufficiency of public hospitals. In terms of the spatial distribution of public hospitals in Chongming, all public hospitals are located on the main island, and there are no public hospitals on the other two smaller islands (i.e., Changxing Island and Hengsha Island), causing a marked accessibility disparity within the district. As a designated eco-island, however, the vision toward ecodevelopment and a green economy might limit the number of residential neighborhoods and intensity of development. Even though the challenge of imbalanced spatial distribution of public hospitals exists in multiple districts, each must be examined on a case-by-case basis.

In the district-level analysis of accessibility to public hospitals, there are two groups of districts that share similar population, acreage, and the number, class structure, and density of public hospitals, but the accessibility level of one district is markedly better than the other. The more balanced spatial distribution of public hospitals of the former district is the main reason for the difference. Detailed information on the two pairs was extracted from the preceding results and summarized in Tables 3 and 4 and Appendix B.

Table 3 The public hospital information of Jiading and Songjiang

District	Population (%)	Acreage (%)	Public hospital			Public hospital density (place/km²)	Occupancy of public hospital (place/100,000 people)
			Second class	Third class	Total
Jiading	6.56	7.32	5	2	7	0.0151	0.44
Songjiang	7.27	9.55	6	2	8	0.0132	0.45

Table 4 The public hospital information of Jinshan and Fengxian

District	Population (%)	Acreage (%)	Public hospital			Public hospital density (place/km²)	Occupancy of public hospital (place/thousand people)
			Second class	Third class	Total
Jinshan	3.32	9.24	4	2	6	0.0102	0.0075
Fengxian	4.75	10.84	4	1	5	0.0073	0.0043

Conclusions

This study focuses on the spatial distribution of public hospitals in Shanghai, which is analyzed through the spatial patterns and the accessibility indicators calculated based on the improved gravity model and the travel time generated by the Baidu Web mapping API. The main results of this study are as follows: (1) Public hospitals in Shanghai, especially the third-class facilities with better health care services, are mainly concentrated in the central urban area, although the population and acreage of the central urban area are relatively small. The population density disparity is also larger in the central urban area, however. (2) The accessibility to public hospitals gradually decreases from the central urban area to the noncentral area, although there are hotspots of high accessibility in each noncentral district.

Compared with existing studies, this study closes research gaps by integrating the big data–based Web mapping API into an improved gravity model that considers the class, capacity, and nearby population’s competition for public hospitals to improve the precision of accessibility analysis without sacrificing comprehensiveness. Such methodology has been seldom seen in relevant studies, especially in the Chinese context. Also, this study follows a sensitivity analysis approach in the accessibility analysis from city level to district level. For example, in the city-level analysis, the relatively poor accessibility of Changning to its central urban district peers is not evident because patients in Changning can also go to other nearby hospitals in nearby districts. In the district-level analysis, however, its lack of high-level hospitals is clearer because the assumption is that patients in Changning can only go to hospitals in Changning. As we addressed in the Discussion, this can cause problems, such as putting extra pressure on hospitals in nearby districts and also resucing the accessibility for patients in Changning.

This study has some limitations, however, and needs to be further improved⁷ in the following four ways. First, confined by the availability of the quantity and distribution of population, the smallest research unit is the subdistrict when calculating the accessibility indicator in this study, and it is hypothesized that all residents in the same subdistrict depart from the same subdistrict government office, which is not necessarily the geographic center of the subdistrict. Second, this study does not compare the results with land use patterns of Shanghai. Land use plays a great part, however, in deciding the location of public hospitals. Third, this study only includes public hospitals, but private hospitals are an important part of the health care system and an important choice for patients when seeking medical treatment. Thus, the planning of public hospitals will also take the existing private hospitals into consideration. Therefore, only focusing on the spatial distribution balance of public hospitals cannot accurately reflect the rationale of health care resource planning in Shanghai. Fourth, all of the analyses in this study are quantitative, and the situations on the ground are estimated based on points through the interpolation model with no concrete case or example for support.

Supplemental Material

Supplemental data for this article can be accessed on the publisher’s site at: http://dx.doi.org/10.1080/00330124.2021.2000445. Data and code used in this article can be found at https://drive.google.com/drive/folders/1HgpTe-eqmfExn3kFSVGKghY-MZSSmFUX?usp=sharing. The data that support the findings of this study are openly available in figshare at https://doi.org/10.6084/m9.figshare.16544973.

Additional information

Funding

This research was funded by the DURF grant and the Zaanheh Project fund of NYU Shanghai, sponsored by the Open Project Fund of the Shanghai Key Lab for Urban Ecological Processes and Eco-Restoration Grant No. 2020SHUES0A01, Key Program of the National Natural Science Foundation of China Grant No. 7183000103, and the Shanghai Pujiang Program Grant No. 2019PJC076. We received support from the PEAK Urban programme of University of Oxford, funded by UKRI’s Global Challenge Research Fund Grant Ref: ES/P01105 5/1.

Notes on contributors

Yiling Shi

YILING SHI is an Undergraduate Student in Social Science (Urban Studies track) and Business & Finance at New York University Shanghai, Shanghai 200122, China. E-mail: [email protected]. Her professional and academic interest lies at the intersection of urban planning, economic development, and public policy in the Chinese context.

Junyan Yang

JUNYAN YANG is a Professor in the Department of Urban Planning, School of Architecture, Southeast University, Nanjing 210096, China. E-mail: [email protected]. He is also the Director of the Urban Center Research Institute at Southeast University. His research focuses on smart city big data, from basic theory to key technology to engineering practice and big data forecasting and smart city design.

Michael Keith

MICHAEL KEITH is a Professor at COMPAS, University of Oxford, Oxford OX2 6QS, UK. E-mail: [email protected]. He is the Director of the PEAK Urban Research program, Coordinator of Urban Transformations (The ESRC portfolio of investments and research on cities), and Codirector of the Oxford Programme for the Future of Cities. His research focuses on migration-related processes of urban change.

Kun Song

KUN SONG is an Assistant Professor at School of Ecological and Environmental Sciences at East China Normal University, Shanghai, China 200241. E-mail: [email protected]. His research interests include urban biodiversity conservation and ecological restoration, and urban vegetation mapping.

Ying Li

YING LI is an Assistant Professor of Practice and Codirector of the Shanghai Urban Lab at New York University Shanghai, Shanghai, China 200133. E-mail: [email protected]. Her research interests focus on the emerging challenges and opportunities in the process of urbanization. Her expertise lies in urban renovation of historical districts in Chinese cities.

ChengHe Guan

CHENGHE GUAN is an Assistant Professor of Urban Science and Policy at New York University Shanghai, Shanghai, China 200122. E-mail: [email protected] His research interests include urban big data and space–time computation on digital smart cities in a multidisciplinary context and sustainable urban planning and simulation of low-carbon cities from a human mobility perspective.

Notes

1 Shanghai is one of the four provincial-level municipalities in China. With an area of 6,340.50 km², Shanghai is located at the mouth of the Yangtze River in eastern China. Shanghai is regarded as one of the most developed cities in China with a population of 24.24 million, a gross domestic product of 3.27 trillion RMB (US$467 billion), and an urbanization rate of 88.1 percent at the end of 2018 (Shanghai Bureau of Statistics 2019). The Annual Report on China’s Hospitals Competitiveness (2018–2019) provides a Top 100 chart of Chinese hospitals, in which, geographically speaking, Shanghai ranks second only to the capital city of Beijing (Zhuang et al. 2019). By the end of 2018, there were 364 public and private hospitals and 246 community health care centers in Shanghai (NHCPRC 2019). With 139,029 patient beds in total, the average patient bed number per thousand people in Shanghai was only 5.74 in 2018, below the national average of 6.03 and tied for twentieth among thirty-one provinces with Beijing. It is noteworthy that although the eastern China region included quantitatively more hospitals and had a higher portion of high-level hospitals than the middle and the western regions, the average patient bed number per thousand people in the eastern region was the lowest among the three at 5.60 (NHCPRC 2019). Therefore, Shanghai is a fair representative of the eastern provinces facing the allocation problem of medical resources to the mass population. This study covers all of the 166 public hospitals located in Shanghai, including all of the branch public hospitals. The Evaluation Methods for Medical Institutions of China categorized all health institutions into third-, second-, and first-class mainly on the basis of the health care service level and the scale of the hospital, with third-class the best and first-class the least; each class is further classified into three subclasses—A, B, and C—with A the best and C the least (NHCPRC 2018). In the public health system of Shanghai, there are only five available classes. All public hospitals are either third-class (3A or 3B) or second-class (2A or 2B). The first-class public health institutions, with no further subclassification, are community health care centers that are not classified as public hospitals by Shanghai Regional Health Planning (2011–2020; SMPG 2010). The location, number of patient beds, and class of the public hospitals are obtained from the Shanghai Healthcare Service Searching System Web site sponsored by the Shanghai Municipal Health Commission (see http://jg.soyi.sh.cn/). Table B.2 provides a summary of the public hospitals. For branch hospitals without class and patient bed data, the class is set the same as the corresponding parent hospital, and the patient bed data are obtained from the official Web site of the corresponding parent hospital. Among the 166 public hospitals, 58 are 3A public hospitals, 16 are 3B public hospitals, 65 are 2A public hospitals, and 27 are 2B public hospitals.

2 When a route is requested between a starting point and a destination in Baidu Map, the system can come up with an ideal route, and the time taken for this route based on real-time traffic conditions. To eliminate all possible errors, prewritten code (Code S2) was adopted in Python that (1) inserts a Microsoft Excel file of all route requests with the coordinates of the starting position and the destination (coordinates are generated by the geocoding module of Baidu’s Web mapping API [Code S1]), (2) sends all of the requests to the route planning module of Baidu’s Web mapping API and runs, and (3) downloads all of the resulting data and saves them in a new Excel file.

3 The survey shows that third-class public hospitals are generally larger in scale and better in health care provision, so they can attract potential patients despite the distance. Second-level public hospitals are generally smaller in scale and limited in health care provision, so they are not attractive to distant patients and the average maximum time patients would like to spend traveling is thirty minutes. The accessibility indicators of all subdistricts are imported into ArcGIS and hierarchically displayed on the maps with the inverse distance weighting (IDW) analysis. IDW is a deterministic method for multivariate interpolation with some known points and works by assigning weighted values to unknown points based on available values at the known points. IDW assumes that points decrease in influence with distance from the measured location. The higher the accessibility indicator, the better accessibility level of the area.

4 The average public hospital density of Shanghai is 0.0262/km². The density in the central urban area is more than ten times higher than the Shanghai average, whereas the density in the noncentral area is half of the Shanghai average. Among the central districts, Huangpu has the highest density (0.6843/km²) and Putuo has the lowest (0.1277/km²). Among the suburban districts, Baoshan has the highest density (0.0369/km²) and Chongming the lowest (0.0051/km²). The average occupancy of public hospitals per thousand people in Shanghai is 0.0068. Huangpu has the highest occupancy level of (0.0214) and Qingpu has the lowest (0.0033).

5 The third-class public hospitals with wide coverage also account for a relatively high proportion in these districts. The overlapping of multiple wide coverage areas brings about the spillover effect of high-quality public health care services, effectively improving the accessibility level. As a result, even though the spatial distribution of public hospitals is not well balanced, the accessibility indicator in more than half of the area exceeds 350, the threshold for high accessibility, in these three districts. The overall average of central urban districts is 379, higher than 350, the floor boundary of the highest section. Huangpu District has the highest accessibility level and is at the center of the central urban area, but other central urban districts surrounding it generally present a descending pattern of accessibility. The accessibility levels of Hongkou and Jing’an are in the middle-high section with indicator values mainly between 250 and 350. In these two districts, the accessibility level gradually decreases from the area close to the city center to the area far away from the city center. Although the number and class composition of public hospitals in these two districts are different, they both concentrate in the area closer to the city center, leading to a significant accessibility disparity. Putuo and Changning are the two districts with the lowest accessibility level in the central urban area with indicator values mainly in the middle range (150–250). The numbers of third-class public hospitals in these two districts are the smallest among the central urban area, causing the sparse distribution patterns of the public hospitals. Consequently, the overall accessibility level is low. On the other hand, the disparity is high. There are some areas with very low accessibility levels in the low-middle section (75–150) far away from the city center, such as the northwest of Putuo and the northwest and the southwest of Changning.

6 Minhang, Jiading, Songjiang, and Jinshan have relatively higher accessibility levels within the suburban area, which is mainly in the middle section (150–250). In Minhang, the total number of third-class public hospitals is the highest in the suburban area, and the spatial distribution of public hospitals is relatively balanced. Although the number of public hospitals is limited in Jiading, Songjiang, and Jinshan, they exhibit different spatial arrangements of public hospitals. For example, the public hospitals in Jiading and Songjiang are located in the geographic center so that the accessibility levels in the districts are relatively balanced and gradually decrease from the center to the periphery. Moreover, Jiading allocates some public hospitals in the western part, the farthest away from the city center, to alleviate the imbalance of accessibility. Jinshan is in a flat shape in the east–west direction, and the public hospitals are distributed along the border of the district. Therefore, the accessibility level of Jinshan has very low disparity. Fengxian, Baoshan, Qingpu, and Chongming have a low accessibility level, which is mainly in the low and low-middle range (9.7–150). The limited third-class public hospitals lead to the overall low level of accessibility in Baoshan, and the concentration of public hospitals in the central and the southern sections causes extremely low accessibility level in the northern area. With large geographic areas to cover, third-class public hospitals in Fengxian, Qingpu, and Chongming are inadequate, and the spatial distribution of the existing public hospitals is not balanced, causing low accessibility levels and high disparity in these districts.

7 In future studies, the smallest research unit can be set at the community level, the administrative unit one level smaller than the subdistrict. The quantity and distribution of population at the community level can further improve the accuracy of density distribution and accessibility analysis. Also, land use will be included, not only for a more accurate and comprehensive analysis of the public hospital spatial distribution pattern but also as a crucial factor to offer optimizing suggestions that are more feasible and practical. Moreover, future studies should include private hospitals as well as community services in the study scope and offer more comprehensive suggestions for optimizing the health care system. Finally, a comparison between the estimated situation and the actual situation should be added to ensure the accuracy of models. Qualitative research methods such as interviews and field trips can be supplemented to validate the results and reflect user experiences. Also, we plan to test sensitivity of different weighting systems, apply location-based digital devices to reveal temporal variation of accessibility to public hospitals, and consider the emergent telemedicine and digital health care delivery online services to provide a more comprehensive evaluation.

Appendix A:

The District-Level Accessibility to Public Hospitals of Shanghai

Figure A1 Accessibility of Jing’an, Yangpu, Hongkou, Putuo, Changning, and Pudong.

Figure A2 Accessibility of Jiading, Minhang, Songjiang, Jinshan, Fengxian, Baoshan, Qingpu, and Chongming.

The first group is Jiading and Songjiang. Their occupancy of public hospitals per thousand people values are similar, but the accessibility level of Jiading is higher than that of Songjiang. In addition to deploying public hospitals in the geographical center of the district like Songjiang, Jiading deploys three public hospitals in the southwest area with a large population. Anting Town in southwestern Jiading has the second highest population density of roughly 1,213 people/km² among the subdistricts on the periphery of Jiading. In Jiading, this area is the farthest away from the city center and thus has the greatest demand for accessible public hospitals in the area. Strengthening the accessibility level of this area can effectively reduce the accessibility disparity and thus improve the average level of accessibility.

The second group is Jinshan and Fengxian. With a similar population, the occupancy of public hospitals per thousand people in Jinshan is higher than that in Fengxian, and the accessibility level of Jinshan is also higher than that of Fengxian. Jinshan decides the deployment of public hospitals based on the geographic shape of the district instead of the common strategy that deploys public hospitals in the geographic center. The geographic shape of Jinshan is narrow from east to west and its eastern area is comparatively wider from north to south. Based on this shape, the public hospitals in Jinshan are basically evenly deployed along the border, with three public hospitals on the north border, east border, and northwest–southeast border if double-counted. There is no point in the district that is significantly farther away from public hospitals than other points. Therefore, the average accessibility level of Jinshan is high with limited disparity.

Appendix B:

Statistical Information

 

 

Table B.1 Patient beds per 1,000 population in Mainland China

Year region/province	Patient beds			Patient beds per 1,000 population			Patient beds per 1,000 rural population
	Total	Urban	Rural	Total	Urban	Rural
2010	4,786,831	2,302,297	2,484,534	3.58	5.94	2.60	1.04
2013	6,181,891	2,948,465	3,233,426	4.55	7.36	3.35	1.18
2014	6,601,214	3,169,880	3,431,334	4.85	7.84	3.54	1.20
2015	7,015,214	3,418,194	3,597,020	5.11	8.27	3.71	1.24
2016	7,410,453	3,654,956	3,755,497	5.37	8.41	3.91	1.27
2017	7,940,252	3,922,024	4,018,228	5.72	8.75	4.19	1.35
2018	8,404,088	4,141,427	4,262,661	6.03	8.70	4.56	1.43
Eastern Region	3,253,527	1,920,520	1,333,007	5.60	8.44	4.28	1.26
Middle Region	2,687,513	1,191,760	1,495,753	6.17	9.87	4.25	1.37
Western Region	2,463,048	1,029,147	1,433,901	6.49	8.92	4.89	1.57
Beijing	123,626	123,626	—	5.74	8.69	—	—
Tianjin	68,247	64,091	4,156	4.37	6.20	6.12	6.12
Hebei	421,916	172,918	248,998	5.58	7.53	4.51	1.30
Shanxi	208,305	102,839	105,466	5.60	10.26	4.18	1.23
Neimenggu	159,006	78,117	80,889	6.27	11.30	4.66	1.29
Liaoning	314,440	207,897	106,543	7.21	10.45	4.86	1.46
Jilin	166,994	86,256	80,738	6.18	10.19	4.64	1.04
Heilongjiang	250,129	158,359	91,770	6.63	12.13	4.12	1.07
Shanghai	139,029	135,963	3,066	5.74	9.72	4.51	—
Jiangsu	491,522	272,211	219,311	6.11	8.25	4.80	1.55
Zhejiang	332,086	186,155	145,931	5.79	9.11	4.84	0.62
Anhui	328,123	156,568	171,555	5.19	7.27	3.46	1.18
Fujian	192,473	89,718	102,755	4.88	7.25	3.85	1.16
Jiangxi	249,490	97,089	152,401	5.37	7.96	3.97	1.47
Shandong	608,459	289,209	319,250	6.06	8.20	4.79	1.46
Henan	608,519	235,146	373,373	6.34	10.83	4.01	1.23
Hubei	393,514	182,653	210,861	6.65	9.13	5.05	1.86
Hunan	482,439	172,850	309,589	6.99	12.10	5.24	1.72
Guangdong	516,929	355,018	161,911	4.56	7.38	3.35	1.25
Guangxi	255,940	110,629	145,311	5.20	6.38	3.66	1.65
Hainan	44,800	23,714	21,086	4.80	9.47	3.14	0.90
Chongqing	220,104	129,265	90,839	7.10	7.47	5.41	2.50
Sichuan	598,898	255,095	343,803	7.18	8.43	5.64	2.16
Guizhou	245,639	76,734	168,905	6.82	8.60	4.55	1.14
Yunnan	291,194	81,459	209,735	6.03	10.36	5.23	1.32
Xizang	16,787	8,442	8,345	4.88	7.01	3.48	1.52
Shaanxi	253,711	133,165	120,546	6.57	8.21	4.97	1.49
Gansu	162,737	72,912	89,825	6.17	8.74	4.59	1.37
Qinghai	39,146	17,853	21,293	6.49	11.09	4.92	1.05
Ningxia	41,005	25,908	15,097	5.96	8.12	4.14	0.97
Xinjiang	178,881	39,568	139,313	7.19	11.30	7.12	1.51

Note: The population for patient beds per 1,000 people in total is permanent residential population; the populations for patient beds per 1,000 people for urban area and rural area are calculated based on registered population.