Is Urbanization Good for the Health of Middle-Aged and Elderly People in China?—Based on CHARLS Data

Abstract

The purpose of this paper is to test whether improved healthcare services can mitigate health hazards resulting from environmental pollution in the urbanization process. Specifically, using China Health and Retirement Longitudinal Study (CHARLS) data and official statistics, this paper constructs comprehensive urbanization indicators and healthcare service indicators by applying the fully arrayed polygonal graphical indication method. Then, we introduce healthcare service indicators, urbanization indicators, environmental pollution indicators, and the interaction term between environmental pollution and healthcare into an ordered-logistics regression model. Our results indicate that improvement in health conditions can decrease the health risks from multiplied emissions of industrial sulfur dioxide, industrial soot and dust, and industrial effluents, but it cannot counteract the adverse health effects of PM_2.5. Furthermore, heterogeneity tests show that, when considering the multidimensionality of urbanization, the positive influence of healthcare is the greatest in residential surroundings urbanization and economic urbanization, which reduces the prevalence of chronic diseases by 18.4% and 14.9%, respectively. Among the diverse city types, mixed-economy cities have the most obvious positive effects, where healthcare has the greatest mitigating effect on the health damage caused by industrial sulfur dioxide and industrial soot and dust, decreasing the prevalence of chronic diseases among the middle-aged and elderly by 27.3% and 16.4%, respectively. When considering the regional impacts of urbanization, there is a large difference in the positive effects brought about by medical care, which is reflected mainly in eastern and western China. In eastern China, although healthcare does not offset the health damage of PM_2.5, the increase in chronic diseases among the middle-aged and elderly is only 0.5%, while in western China, the increase rises to 22.4%.

1. Introduction

Since the 1990s, both the speed and scale of China’s urbanization have accelerated [1]. During this period, the urbanization rate in China has increased rapidly from 30.48% to 60.60%, an increase of nearly 31%. It has been reported that the urbanization rate in China will reach 75% by 2030 (The China urbanization 2.0 report). According to that result, it can be seen that, in several decades, China will have achieved the urbanization progress that occurred in western countries over hundreds of years [2]. Such rapid urbanization growth poses large challenges to public health. Specifically, quick urbanization leads to serious problems of environmental pollution and unhealthy lifestyles [3]. Among these problems, the most widespread impact has been on air pollution [4], for example, increased levels of industrial sulfur dioxide [5] and particulate matter (PM₁₀, PM_2.5) pollution [6]. PM_2.5 pollution has had a particular impact, according to a previous study; both long-term and short-term exposure to PM_2.5 increased the probability of chronic diseases for residents, such as respiratory disease and cardiovascular morbidity [7]. It is even reported that 48.6% of the Chinese population (nearly 100 million people) suffer from obstructive pulmonary disease (COPD), and 18.7% of COPD deaths are attributable to environmental PM_2.5 exposure (China Pulmonary Health Study). At the same time, the hazards of wastewater to human health cannot be ignored [8]. Current research indicates that water pollution is associated not only with acute waterborne diseases (cholera, diarrhea) [9] but also with cancer risk in severe cases [10]. The poor lifestyles and unhealthy diets resulting from urbanization also introduce many risks for residents’ health [11], such as obesity [12], hypertension [13], and heart disease [14]. Despite the negative impacts of urbanization, it also leads to an improved standard of living, better healthcare resources, increased job opportunities, and higher education levels, which lead to greatly improved public health [6]. However, the most sensitive of the impacted groups are middle-aged and older adults. According to statistics, China’s elderly population accounted for 18.1% of the total population in 2019, and the proportion is expected to rise to 20% by 2030. At that time, China will enter a heavily aging society. Furthermore, with an accelerating rate of increase in the proportion of the population that is aged [15], chronic diseases among the elderly are a particular problem in China [16]. According to statistics, nearly 80% of deaths in the elderly population are caused by chronic diseases, and the elderly are 3.2 times more likely to suffer from chronic diseases than other groups [17], with environmental pollution factors increasing the probability of chronic diseases in the elderly. Researchers have shown that in Wuhan, China, a 10 µg/m³ increase in NO₂ is associated with a 1.6% increase in CVD mortality in the elderly [18]. Some scholars have even predicted the future trend of CVD in China; by 2030, the number of CVD events per year in China will increase by more than 50% due to population aging and population growth alone [19], of which environmental pollution is a factor that cannot be ignored. These statistics demonstrate the importance of studying the health effects of urbanization on middle-aged and elderly people in China. Based on this, are the positive or negative effects of urbanization on the health of older adults greater?

The influence that urbanization has on health is complex [20,21]. Some scholars argue that urbanization adversely affects the health of residents [20,22]. Previous research has considered the negative health effects of single factors such as changes in environmental pollution [3,23], unhealthy lifestyles, and socioeconomic status [22] caused by urbanization. Among these factors, researchers reported that environmental pollution has the greatest influence on health. Other scholars believe that in the process of urbanization rising living standards, improved healthcare resources, more job opportunities, and higher educational levels have dramatically enhanced public health [6,24]. In particular, it should be noted that the advancement of sanitary conditions brought about by urbanization has made it easier for people to enjoy the latest achievements in medical technology and to obtain better medical care, which has benefited city residents and promoted their health. In conclusion, it can be seen that there are disparate impacts of urbanization on health.

In studies of the implications of urbanization for health, most of the scholars have focused on single indexes such as the demographic urbanization rate [25], constructing the urbanization indicators of broad community characteristics [22], nighttime light data [26], and so on. Even when researchers have taken into account the complexity and multidimensionality of urbanization, they have only briefly adopted multiple indexes or have used data that were not representative. For example, Liu et al. [27] considered the effects of differing levels and rates of urbanization on health by using indexes of land-use transitions, the growth of the economy, population clustering, and health services, and Chen et al. [26] adopted nighttime databases to assess the effects of urbanization of different dimensions, development speed, and level of health in county-level regions. However, only some scholars consider the multidimensionality of urbanization and employ the entropy method [28] or the fully arrayed polygonal graphical indication method [29] to divide urbanization into the urbanization of the population, urbanization of the economy, urbanization of residential environments, and urbanization of residential conditions. Although the selected methods and indicators can objectively reflect the urbanization of all of China, most of them are only analyzed at the provincial level and not at the urban level. More importantly, these measurements of urbanization do not apply to the impact of urbanization on health. Therefore, this paper adopts the fully arrayed polygonal graphical indication method to reconstruct the comprehensive urbanization index for China’s prefecture-level cities and incorporate it into a model of urbanization and health, as well as dividing it into four aspects: demographic urbanization, economy urbanization, residential surroundings urbanization, and residential condition urbanization.

Most scholars studying healthcare service indicators use single indexes, such as healthcare expenditures [30], the number of hospital beds [31,32], and the number of doctors in hospitals [33]. However, medical resources are not only a reflection of these singular indexes, but are also closely related to financial resources, material resources, and manpower for healthcare. Therefore, some researchers have employed the entropy method [34] to comprehensively evaluate healthcare resources, but these methods are mostly applied at the provincial scale in China; the data are not easily available at the municipal level. Thus, taking into account the availability of data and the applicability of the method, this paper employs the fully arrayed polygonal graphical indication method [29] to comprehensively evaluate the level of medical and health resources by selecting three indicators: the number of hospital beds per thousand people, the number of medical and health institutions per thousand people, and the number of hospitals per hundred square kilometers.

To summarize, this paper adopts the fully arrayed polygonal graphical indication method to calculate composite urbanization indexes and healthcare indicators. In this method, we construct the comprehensive urbanization indicators based on the four dimensions of demographic urbanization, economy urbanization, residential surroundings urbanization, and residential condition urbanization. Next, the urbanization indicators, the health service indicators, and the interaction terms of environmental pollution and health services are incorporated into an ordered logistic model to explore the advantages and disadvantages that change in healthcare conditions and environmental pollution caused by urbanization and see what effect they have on the health of middle-aged and elderly people. The heterogeneity of urbanization in multidimensionality, regional differences, and city types are considered as well. The primary contributions of this paper are as follows: First, we incorporate urbanization, environmental pollution, healthcare services, and health into the same research framework to analyze the dual effects and the magnitude of the impacts of environmental pollution and the improvement of healthcare service caused by urbanization. Then, the comprehensive urbanization and healthcare service indexes measured by the fully arrayed polygonal graphical indication method are integrated into an ordered logistics model of the impact of urbanization on health, and the interaction term between the environmental pollution and healthcare service is introduced to study the negative and positive impacts of urbanization on health and the extent of those effects. Finally, the comprehensive urbanization indicators calculated by the fully arrayed polygonal graphical indication method are used to divide the integrated urbanization into four dimensions: urbanization of the population, urbanization of the economy, urbanization of residential environment, and urbanization of residential conditions. This is then used to explore the heterogeneity of the influence of urbanization on health based on the different dimensions of urbanization, the regional variability of urbanization, and the variability of city types. The manuscript is arranged as follows: the second part contains the data and methods, the third part contains the results and discussion, and the last part contains the conclusion and policy implications.

2. Methods and Data

2.1. Methods

2.1.1. Constructing the Comprehensive Indicator System of Urbanization

Most scholars apply the rate of demographic urbanization to calculate the urbanization level [25]. However, urbanization is a complex concept with a variety of dimensions and characteristics [20], including not only demographic aspects but also economic, cultural, social, and ecosystem aspects. Thus, the comprehensive index approach can better demonstrate the integrative characteristics of urbanization. This study reconstructs the consolidated indicator system for measuring urbanization based on the research of Wu et al. [28,29]. The index system can be classified into three layers: The first layer is the target level, that is, the comprehensive urbanization level. The second layer is the criterion level, which contains the indicators demographic urbanization, economy urbanization, residential surroundings urbanization, and residential condition urbanization. The third layer is the indices level, which covers 16 sub-indicators (as shown in

Table 1. The comprehensive urbanization indicator system.

Target Level	Criterion Level	Indicator Level	Unit	Indicator Attribute
Comprehensive urbanization	Demographic urbanization	City Population Density $\left(X_1\right)$	person/square kilometer	rising
		Percentage of urban population$\left(X_2\right)$	%	rising
		Percentage of tertiary sector employees $\left(X_3\right)$	%	rising
	Economy urbanization	Per capita GDP $\left(X_4\right)$	Yuan	rising
		Per capita disposable income of urban residents $\left(X_5\right)$	Yuan	rising
		The tertiary sector as a proportion of GDP $\left(X_6\right)$	%	rising
	Residential surroundings urbanization	The capacity of daily sewage treatment in the city $\left(X_7\right)$	Million cubic meters/day	constrained
		Green coverage rate in the built-up area$\left(X_8\right)$	%	constrained
		Per capita park green area$\left(X_9\right)$	Square meter	constrained
		Harmless treatment rate of domestic waste (the whole city)$\left(X_{10}\right)$	%	constrained
		The integrated utilization rate of general solid waste$\left(X_{11}\right)$	%	constrained
	Residential condition urbanization	Urban water coverage rate$\left(X_{12}\right)$	%	rising
		Urban gas penetration rate$\left(X_{13}\right)$	%	rising
		Number of toilets per 1000 population$\left(X_{14}\right)$	unit	rising
		Number of buses per 1000 population (city district)$\left(X_{15}\right)$	unit	rising
		Per capita urban road space$\left(X_{16}\right)$	Square meter	rising

Note: The selected indicators are categorized as rising or constrained based on their attributes. Of these, the rising indicators represent positive contributions of the integrated level of urbanization, while the constrained indicators represent the environmental cost of integrated urbanization development.

). We measure the integrated urbanization layer using the fully aligned polygonal graphical indexing method. Table 1 presents detailed information on the comprehensive urbanization indicator system.

2.1.2. The Calculating Methods of the Comprehensive Urbanization Indicator System

As described in Section 2.1.1, the integrated urbanization approach employs the fully aligned polygonal graphical indexing method to deal with the composite urbanization in China.

The fundamental principle of the method is that N indicators are set up and the higher limit value of these indexes is taken as the radius to construct a central N-side shape. In other words, the connecting line of each index value is used to construct an uneven central N-polygon. The vertex of the n-index is the full permutation of N indexes, which can form (n − 1)!/2 distinct non-regular central n-squares. The complex indices are specified as the ratio of the mean area of all these non-regular polygons to the area of the central polygon. The specific calculation process is as follows:

(1)

The hyperbolic standardization function is applied to standardize the index values:

$$F\left(x\right)=\frac{a}{bx+c}$$

(1)

where a, b, and c represent the parameters of the hyperbolic function, respectively.

(2)

The hyperbolic standardized function fulfills F(U) = 1, F(T) = 0, F(L) = −1, where U is the higher limit of the exponent X, L is the lower limit of X, and T is the critical value of the X. The threshold value can be expressed as the exponential mean.

$$F\left(x\right)=\frac{\left(U-l\right)\left(x-T\right)}{\left(U+L-2T\right)x+UT+LT-2UL}$$

(2)

From Equation (2), the standardized function, F(x), maps the index value located in [L, U] to [−1, 1]. The normalization procedure will lead to a fast-slow-fast nonlinear growth trend of the indicator values.

(3)

For the ith index, the single index value, S_i, is:

$$S_i\left(x\right)=\frac{\left(U_i-l_i\right)\left(x_i-T_i\right)}{\left(U_i+L_i-2T_i\right)x+U_i{T}_i+L_i{T}_i-2L_i{U}_i}$$

(3)

The vertex of the nth-edge consists of the value at $S_i$= 1, the hub point consists of the value at$S_i$= −1, and the threshold value of the polygon index is the value at $S_i$= 0. If the indicator value is above the threshold value, the value of each indicator is positive; otherwise, it is negative.

(4)

The fully aligned polygonal graphical indexing method is indicated as follows:

$$S\left(x\right)=\frac{{{\displaystyle \sum }}_{i\ne j}^{i,j}(S_i+1)\left(S_j+1\right)}{2n\left(n-1\right)}$$

(4)

2.1.3. The Model of the Relationship between Urbanization and Health

To explore the dual effects that the factors of improved medical care service and increased environmental pollution from urbanization have on chronic diseases in middle-aged and elderly people, and considering that the dependent variable is a discrete order variable, this research applies the ordered-logistics model. The constructed baseline model is as follows:

$$Healt{h}_i=\alpha +{\beta }_{1}Urban+{\beta }_{2}Mhs+{\beta }_{3}A{P}_j+\lambda X_i+{\delta }_i$$

(5)

In this formula, H represents the disease status of the micro individual; Urban is the urbanization index, including comprehensive urbanization (CU), demographic urbanization (DU), economy urbanization (EU), residential surroundings urbanization (RSU), and residential condition urbanization (RCU); $A{P}_j$ represents environment pollutants (industrial sulfur dioxide, industrial soot and dust, industrial wastewater) and the yearly average concentration of PM_2.5; $Mhs$ is the medical and health services index; $X_i$ is the vector of the individual-level control variables, including age, sex, education, material, etc.; ${\delta }_i$ is a random perturbation term; and ${\beta }_1$, ${\beta }_2$, and ${\beta }_3$represent the effects of urbanization, medical services, and environmental pollution, respectively, on the health status of the micro individual.

To further study the effect of environmental pollution and healthcare service on the health of middle-aged and elderly people, and consider which is more important, this paper introduces an interaction term of environmental pollution and medical health service to construct the following model:

$$Healt{h}_i=\alpha +{\beta }_{1}Urban+{\beta }_{2}Mhs+{\beta }_{3}A{P}_j+{\beta }_{4}A{P}_j\times Mhs+\lambda X_i+{\delta }_i$$

(6)

In the formula, $A{P}_j\times Mhs$ is the interaction term of environmental pollution (industrial sulfur dioxide, industrial soot and dust, industrial wastewater) and medical care service.

To research the interaction term using the ordered-logistics model, our study calculates the partial effect models of environmental pollution and medical care service on the health status of middle-aged and elderly people to specify the magnitude of this effect. The specific bias effects model is as follows:

$$\frac{\partial P_i}{\partial A{P}_j}=\left[𝓰\left({\mu }_{j-1}-X_{\alpha }^{\prime }\right)-𝓰\left({\mu }_j-X_{\alpha }^{\prime }\right)\right]{\beta }_1+\left[𝓰\left({\mu }_{j-1}-X_{\alpha }^{\prime }\right)-𝓰\left({\mu }_j-X_{\alpha }^{\prime }\right)\right]{\beta }_{4}Mhs$$

(7)

$$\frac{\partial P_i}{\partial Mhs}=\left[𝓰\left({\mu }_{j-1}-X_{\alpha }^{\prime }\right)-𝓰\left({\mu }_j-X_{\alpha }^{\prime }\right)\right]{\beta }_2+\left[𝓰\left({\mu }_{j-1}-X_{\alpha }^{\prime }\right)-𝓰\left({\mu }_j-X_{\alpha }^{\prime }\right)\right]{\beta }_{4}A{P}_j$$

(8)

where, $P_i$ is the probability of the individual health rank and $𝓰\left({\mathsf{\mu }}_{\mathrm{j}-1}-{\mathrm{X}}_{\mathsf{\alpha }}^{\prime }\right)-𝓰\left({\mathsf{\mu }}_{\mathrm{j}}-{\mathrm{X}}_{\mathsf{\alpha }}^{\prime }\right){\beta }_i$ are the partial effects of the ${\beta }_i$ transformation. In the positive and negative examination of $\left[𝓰\left({\mathsf{\mu }}_{\mathrm{j}-1}-{\mathrm{X}}_{\mathsf{\alpha }}^{\prime }\right)-𝓰\left({\mathsf{\mu }}_{\mathrm{j}}-{\mathrm{X}}_{\mathsf{\alpha }}^{\prime }\right)\right]{\beta }_4$, if the value is above zero, it indicates that the improved healthcare conditions cannot counter the adverse effects of environmental pollution on middle-aged and elderly person health, while if it is less than zero, it shows that the improved healthcare service can reduce the negative effects of environmental pollution.

2.2. Variables and Data

2.2.1. Variables

As described in the above model and analysis, the core explanatory variables in the paper are urbanization, medical and health service, and environmental pollution. For the urbanization indicator, the fully arrayed polygonal graphical indication method is used to calculate comprehensive urbanization. Environmental pollution is defined based on per capita industrial wastewater emissions; per capita industrial sulfur dioxide emissions; per capita industrial soot and dust emissions; and the annual average concentration of PM_2.5 [35] because it is not only closely related to economic activities such as industrial structure [36], energy consumption [37], and international trade [38] but is also a key factor in public health [39]. More importantly, in order to maintain the original features of the data and to reduce or eliminate heteroscedasticity in the data, the indicators are presented in their natural logarithms. The healthcare service level is measured using the fully arrayed polygonal graphical indication method for three indicators: the number of beds per thousand people, the number of healthcare institutions per thousand people, and the number of hospitals per hundred square kilometers. In addition, the individual characteristic variables are introduced as control variables, including age, sex, per capita household income, smoking, alcohol use, material, residence, education, and health insurance (as shown in

Table 2. The meaning of variables.

Variable	Definition
Composite urbanization (CU)	Calculated based on demographic urbanization economy urbanization, residential surroundings urbanization, and residential condition urbanization.
Demographic urbanization (DU)	Calculated using the fully arrayed polygonal graphical indication method
Economy urbanization (EU)
Residential surroundings urbanization (RSU)
Residential condition urbanization (RCU)
Medical and health service (Mhs)
Chronic diseases	A person does not have chronic diseases = 1; has contracted 1 or 2 chronic diseases = 2; has contracted 3 or more chronic diseases = 3
Household income per capita	Gross household income/household size (Yuan/person)
Industrial sulfur dioxide	Urban per capita industrial SO2 emissions in 2011, 2013, and 2015 (tons per person)
Industrial soot	Urban per capita industrial soot emissions in 2011, 2013, and 2015 (tons per person)
Industrial wastewater	Urban per capita industrial wastewater emissions in 2011, 2013, and 2015 (tons per person)
PM2.5	The urban annual average concentration of PM2.5 in 2011, 2013, and 2015 (µg/m3)
Education	The highest education: 0 to 6 (Private schools, illiteracy, non-primary school graduates = 0; Primary school graduates = 1; Junior high school graduates = 2; High school graduates = 3; Secondary technical school graduates = 4; College and undergraduate degrees = 5; Master’s degree and above = 6)
Residence	Rural = 1; urban = 0
Age	Age of respondents in 2011, 2013, and 2015
Sex	Male = 1; female = 0
Marriage	Married = 1; otherwise = 0
Healthcare insurance	Obtained healthcare insurance = 1, otherwise = 0
Smoking	More than 100 cigarettes in a lifetime = 1, less than 100 cigarettes = 0
Alcohol use	Consumption of alcoholic beverages, such as beer, wine, or liquor, in the past year, yes = 1, no = 0

).

2.2.2. Data

The data used for the urbanization and environmental pollution indexes originates from The China City Statistical Yearbook, The China Urban Construction Statistical Yearbook, the National Bureau of Statistics of China, and the Provincial Bureau of Statistics. The health and individual characteristic variables are derived from data in the China Health and Retirement Longitudinal Study (CHARLS) [40], which began in 2011 and was repeated every two years and was only updated to 2015. The data primarily focused on China’s middle-aged and elderly over 45 years of age. The study covered a wide range of topics, including information on individual backgrounds, households, health, healthcare and insurance, work, retirement, pensions, income, and basic community information, etc. More information about the CHARLS can be found in Zhao et al. [41]. The summarized data can help meet the needs of scientific research.

Table 3. Descriptive statistics of variables.

Variable	Observation	Mean	S.D.	Min	Max
Chronic disease	21,298	1.914	0.705	1	3
Comprehensive urbanization	21,298	0.384	0.048	0.299	0.545
Demographic urbanization	21,298	0.229	0.084	0.071	0.618
Economy urbanization	21,298	0.230	0.101	0.100	0.619
Residential surroundings urbanization	21,298	0.251	0.111	0.039	0.546
Residential condition urbanization	21,298	0.239	0.083	0.052	0.440
Medical and health service	21,298	0.384	0.048	0.055	0.743
Household income per capita	21,298	8208.774	26,711.500	0.000	3,082,500
Industrial sulfur dioxide	21,298	0.014	0.016	0.000	0.136
Industrial soot and dust	21,298	0.019	0.114	0.000	1.203
PM2.5	21,298	48.252	20.911	3.641	96.540
Industrial wastewater	21,298	16.315	16.515	0.818	181.144
Education	21,298	1.140	1.275	0	6
Age	21,298	61.349	10.141	45	102
Sex	21,298	0.467	0.499	0	1
Marriage	21,298	0.800	0.400	0	1
Residence	21,298	0.696	0.460	0	1
Smoking	21,298	0.564	0.496	0	1
Alcohol use	21,298	0.336	0.472	0	1
Insurance	21,298	0.719	0.449	0	1

Data sources: Data from the China Health and Retirement Longitudinal Study (CHARLS), The China City Statistical Yearbook, The China Urban Construction Statistical Yearbook, the National Bureau of Statistics of China, and the Provincial Bureau of Statistics for 2011, 2013, and 2015. Note: Per capita household income, industrial sulfur dioxide, industrial soot and dust, and industrial wastewater data are in logarithmic form.

reports statistics about variables in the sample. The processing of the data in question is described in detail in the Supplementary Materials.

3. Results and Discussion

This section shows the results of the regression models in estimating the middle-aged and elderly population chronic disease and urbanization, healthcare service, and environmental pollution indexes. First, our paper reports the baseline models; that is, the regression results for health and comprehensive urbanization, healthcare service, and environmental pollution. Second, based on the multidimensionality of the urbanization (demographic urbanism, economy urbanism, residential surroundings urbanism, and residential condition urbanism), the regional difference, and the difference of the city types, the paper explores the heterogeneity on the influence of healthcare service and environmental population on middle-aged and elderly people’s health. Finally, the health multidimensions and the mobility of the population are considered to test the resulting robustness.

3.1. The Influence of Integrated Urbanization on Chronic Diseases in Older Adults

Table 4. The impact of comprehensive urbanization on health.

Variable	Chronic Diseases in Middle-Aged and Elderly People
	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6	Model 7	Model 8
Core Independent Variable
CU	−2.227 *** (0.357)	−2.323 *** (0.357)	−1.602 *** (0.376)	−2.184 *** (0.357)	−2.058 *** (0.360)	−2.491 *** (0.358)	−1.529 *** (0.391)	−2.485 *** (0.359)
Mhs	0.284 * (0.162)	0.336 ** (0.162)	0.242 (0.160)	0.209 (0.161)	0.183 (0.164)	0.451 *** (0.162)	0.236 (0.160)	0.192 (0.159)
ln(PM2.5)	−0.030 (0.025)				−0.009 (0.025)
ln(SO2)		−0.039 *** (0.012)				−0.044 *** (0.012)
ln(Soot)			0.037 *** (0.013)				−0.100 *** (0.018)
ln(Wastewater)				−0.097 *** (0.018)				0.013 (0.013)
Mhs × ln (PM2.5)					0.826 *** (0.243)
Mhs × ln (SO2)						−0.909 *** (0.120)
Mhs × ln (Soot)							−1.114 *** (0.125)
Mhs × ln (Wastewater)								−0.108 (0.160)
Household characteristicsvariables
Household income per capita	−0.019 ** (0.008)	−0.018 ** (0.008)	−0.016 * (0.008)	−0.020 ** (0.008)	−0.019 ** (0.008)	−0.018 ** (0.008)	−0.016 * (0.008)	−0.019 ** (0.008)
Individual characteristicsvariables
Marriage	0.138 *** (0.037)	0.136 *** (0.037)	0.134 *** (0.037)	0.138 *** (0.037)	0.137 *** (0.037)	0.138 *** (0.037)	0.134 *** (0.037)	0.134 *** (0.037)
Smoking	0.043 (0.030)	0.037 (0.030)	0.044 (0.030)	0.047 (0.030)	0.042 (0.030)	0.024 (0.030)	0.044 (0.030)	0.036 (0.030)
Alcohol use	−0.216 *** (0.031)	−0.219 *** (0.031)	−0.219 *** (0.031)	−0.215 *** (0.031)	−0.216 *** (0.031)	−0.223 *** (0.031)	−0.219 *** (0.031)	−0.220 *** (0.031)
Education	−0.011 (0.012)	−0.011 (0.012)	−0.011 (0.012)	−0.014 (0.012)	−0.011 (0.012)	−0.011 (0.012)	−0.011 (0.012)	−0.013 (0.012)
Age	0.034 *** (0.002)	0.034 *** (0.002)	0.034 *** (0.002)	0.034 *** (0.002)	0.034 *** (0.002)	0.034 *** (0.002)	0.034 *** (0.002)	0.034 *** (0.002)
Sex	−0.193 *** (0.032)	−0.189 *** (0.032)	−0.192 *** (0.032)	−0.193 *** (0.032)	−0.191 *** (0.032)	−0.181 *** (0.032)	−0.192 *** (0.032)	−0.183 *** (0.032)
Insurance	0.015 (0.031)	0.020 (0.031)	0.022 (0.031)	0.013 (0.031)	0.017 (0.031)	0.030 (0.031)	0.022 (0.031)	0.020 (0.031)
Residence	−0.143 *** (0.032)	−0.142 *** (0.032)	−0.158 *** (0.032)	−0.146 *** (0.032)	−0.143 *** (0.032)	−0.127 *** (0.032)	−0.158 *** (0.032)	−0.137 *** (0.032)
Number of individuals	21,298	21,298	21,298	21,298	21,298	21,298	21,298	21,298
Number of cities	114	114	114	114	114	114	114	114

Notes: ***, **, and * represent the significance at 1%, 5%, and 10% levels, respectively; the parentheses represent robust standard errors; the income, industrial wastewater, soot and dust, sulfur dioxide, and PM_2.5 values are logarithmic.

reports the model results of Equations (5) and (6). As shown in Table 4, Models 1 to 4 present the effects of per capita industrial sulfur dioxide emissions, per capita industrial soot and dust emissions, per capita industrial wastewater emissions, and PM_2.5 concentrations on chronic diseases in middle-aged and older adults, and Models 5 to 8 introduce the interaction terms of per capita industrial sulfur dioxide emissions, per capita industrial soot and dust emissions, per capita industrial wastewater emissions, and PM_2.5 concentrations with healthcare service to investigate the dual effects of urbanization on health in middle-aged and elderly people. The results of Models 1 to 4 show that integrated urbanization significantly reduces the morbidity of chronic diseases in middle-aged and elderly adults at the statistics level of 1%. As expected, modeling results show industrial soot and dust having a significant negative effect on middle-aged and elderly people, but for industrial sulfur dioxide and wastewater, the results show positive effects for chronic disease, which is in disagreement with objective facts. Thus, the interaction terms were introduced in Models 5 to 8. The estimated results from these models indicate that the coefficients of the interaction terms for industrial sulfur dioxide, industrial soot and dust, and healthcare service are significantly negative, showing that the improvement of healthcare conditions can dramatically relieve the health risks to middle-aged and elderly people caused by increased industrial sulfur dioxide and industrial soot and dust. To depict the effects of healthcare on the mitigation of health hazards posed by environmental pollution, this paper drew marginal effects graphs, shown in Figure 1. The second row of the plots in Figure 1 shows different levels of mitigation effects of healthcare on the health risks produced by industrial sulfur dioxide and industrial soot and dust, with the former decreasing the prevalence of chronic diseases from 0.26 to 0.16 and the latter decreasing prevalence of chronic diseases from 0.31 to 0.16. More noticeably, the coefficient of interaction terms between PM_2.5 and healthcare service is markedly positive, indicating that the enhancement of the healthcare service fails to cancel the health problems from PM_2.5, which increased the prevalence of chronic conditions among the elderly from 0.19 to 0.24 (Figure 1, first row). It is possible that the smaller particle size of PM_2.5 makes it more likely to penetrate through a body’s protective layer than other pollutants and harm public health [42]. However, the interaction term of industrial wastewater and healthcare has a positive effect on chronic diseases in middle-aged and elderly adults, but the effect is not significant.

3.2. The Heterogeneity Analysis of Urbanization on Chronic Diseases in Middle-Aged and Elderly People

According to the baseline analysis described in Section 3.1, comprehensive urbanization can improve the health conditions of middle-aged and elderly people. However, owing to the multidimensionality [26], spatial variability [43], and imbalance [44] in China’s urbanization, there are also significant regional and urban differences in the impact of urbanization on health [45]. Section 3.2 reveals this influence to a greater extent by exploring the heterogeneity of the dual effects of urbanization on health based on the multidimensionality of urbanization, regional differences, and different city types.

3.2.1. The Influence of the Development Trend of Urbanization on Chronic Diseases

Each city has a different direction in the process of urbanization. For example, some cities may focus more on economic development and population clustering in the primary stage of urbanization, while others concentrate more on the improvement of the living environment and living conditions in the mature stage of urbanization. These differences lead to variability in the impact of diverse dimensions of urbanization on health. To address variability, this paper adopted the fully arrayed polygonal graphical indication method to classify the comprehensive urbanization into four dimensions to express the features of the urbanization process: the urbanization of the population, the urbanization of economy, residential surroundings urbanization, and residential condition urbanization. To clearly depict the impact of urbanization on health in different dimensions, this paper not only carried out the corresponding regression (

Table 5. The influence of multidimensionality of urbanization on health.

Variable	Demographic Urbanization				Economy Urbanization
	(1) PM2.5	(2) SO2	(3) Soot	(4) Wastewater	(1) PM2.5	(2) SO2	(3) Soot	(4) Wastewater
DU	−0.909 *** (0.217)	−1.178 *** (0.215)	−0.877 *** (0.222)	−1.096 *** (0.216)
EU					−1.250 *** (0.190)	−1.206 *** (0.183)	−1.499 *** (0.185)	−0.752 *** (0.205)
RSU
RCU
Mhs	−0.003 (0.158)	0.331 ** (0.163)	−0.074 (0.162)	0.377 ** (0.161)	0.466 ** (0.184)	0.484 *** (0.169)	0.420 ** (0.168)	0.274 (0.171)
ln (PM2.5)	0.019 (0.026)				−0.060 ** (0.026)
ln (SO2)		−0.055 *** (0.012)				−0.029 ** (0.012)
ln (Soot)			0.001 (0.014)				0.035 *** (0.013)
ln (Wastewater)				−0.129 *** (0.017)				−0.096 *** (0.019)
Mhs × ln (PM2.5)	0.944 *** (0.241)				0.770 *** (0.243)
Mhs × ln (SO2)		−0.854 *** (0.119)				−0.909 *** (0.119)
Mhs × ln (Soot)			−1.048 *** (0.124)				−1.146 *** (0.125)
Mhs × ln (Wastewater)				−0.055 (0.159)				−0.109 (0.161)
Variable	Residential surroundings urbanization				Residential condition urbanization
	(1) PM2.5	(2) SO2	(3) Soot	(4) Wastew-ater	(1) PM2.5	(2) SO2	(3) Soot	(4) Wastew-ater
DU
EU
RSU	−0.613 *** (0.130)	−0.803 *** (0.129)	−0.750 *** (0.130)	−0.427 *** (0.136)
RCU					−0.119 (0.166)	−0.178 (0.165)	−0.290 * (0.166)	0.166 (0.171)
Mhs	−0.136 (0.141)	0.113 (0.140)	−0.167 (0.138)	0.022 (0.138)	−0.370 *** (0.135)	−0.219 * (0.133)	−0.437 *** (0.132)	−0.157 (0.133)
ln (PM2.5)	0.001 (0.026)				−0.010 (0.025)
ln (SO2)		−0.044 *** (0.012)				−0.038 *** (0.012)
ln (Soot)			0.010 (0.013)				0.022 * (0.013)
ln (Wastewater)				−0.109 *** (0.018)				−0.130 *** (0.018)
Mhs × ln (PM2.5)	0.895 *** (0.242)				0.997 *** (0.242)
Mhs × ln (SO2)		−0.926 *** (0.120)				−0.859 *** (0.119)
Mhs × ln (Soot)			−1.104 *** (0.125)				−1.044 *** (0.124)
Mhs × ln (Wastewater)				−0.198 (0.156)				−0.281 * (0.154)

Notes: ***, **, and * represent the significance at 1%, 5%, and 10% levels, respectively; parentheses represent robust standard errors; the control variables are the same as in Table 1, but the results for the control variables are not reported due to space limitations.

) but also reported the marginal effect of medical service and environmental pollutants on health (

Table A1. The estimation of marginal effects on chronic diseases.

The Multidimensionality of Urbanization	Variable	Marginal Effects
		A Person Does Not Have Chronic Diseases	Contracting 1 or 2 Chronic Diseases	Contracting 3 or More Chronic Diseases
Demographic urbanization	Mhs × ln (PM2.5)	−0.191 *** (0.048)	0.038 *** (0.010)	0.152 *** (0.039)
	Mhs × ln (SO2)	0.172 *** (0.023)	−0.035 *** (0.005)	−0.137 *** (0.018)
	Mhs × ln (Soot)	0.211 *** (0.025)	−0.042 *** (0.005)	−0.169 *** (0.20)
	Mhs × ln (Wastewater)	−0.011 (0.032)	−0.002 (0.006)	−0.009 (0.026)
Economic Urbanization	Mhs × ln (PM2.5)	−0.155 ** (0.049)	0.031 ** (0.010)	0.124 ** (0.039)
	Mhs × ln (SO2)	0.183 *** (0.023)	−0.037 *** (0.005)	−0.146 *** (0.018)
	Mhs × ln (Soot)	0.230 *** (0.025)	−0.046 *** (0.006)	−0.184 *** (0.020)
	Mhs × ln (Wastewater)	0.219 (0.032)	−0.004 (0.007)	−0.018 (0.026)
Residential surroundings urbanization	Mhs × ln (PM2.5)	−0.181 *** (0.049)	0.036 *** (0.010)	0.144 *** (0.039)
	Mhs × ln (SO2)	0.187 *** (0.024)	−0.037 *** (0.005)	−0.149 *** (0.019)
	Mhs × ln (Soot)	0.222 *** (0.025)	−0.044 *** (0.006)	−0.178 *** (0.020)
	Mhs × ln (Wastewater)	0.040 (0.031)	−0.008 (0.006)	−0.032 (0.025)
Residential conditions urbanization	Mhs × ln (PM2.5)	−0.201 *** (0.049)	0.040 *** (0.010)	0.161 *** (0.039)
	Mhs × ln (SO2)	0.173 *** (0.024)	−0.035 *** (0.005)	−0.138 *** (0.019)
	Mhs × ln (Soot)	0.210 *** (0.025)	−0.042 *** (0.005)	−0.168 *** (0.020)
	Mhs × ln (Wastewater)	0.057 * (0.031)	−0.011* (0.006)	−0.045 * (0.025)

Notes: ***, **, and * represent the significance at 1%, 5%, and 10% levels, respectively; parentheses represent robust standard errors.

). The results indicate that demographic urbanization, economy urbanization, and residential surroundings urbanization all significantly reduce the incidence of chronic diseases among elderly people (as shown in Table 5); however, no matter which dimension of urbanization is focused on, improved healthcare fails to offset the health risks caused by increased PM_2.5 from the process of urbanization.

When economy urbanization is the focus dimension, the marginal effects of the interaction term between healthcare and PM_2.5 are the smallest (DU: 0.152; EU: 0.124; RSU: 0.144; RCU: 0.161) (Table A1); that is, when economy urbanization is the main orientation of the urbanization process, the increase in the prevalence of chronic diseases among the middle-aged and elderly adults is the lowest, at 12.4% (Table A1). In addition, in this circumstance, the marginal effects of the interaction between medical care and industrial soot and dust are the largest, which indicates that healthcare has the largest mitigation effect on the health damage caused by industrial soot and dust when economic urbanization is the focus dimension, reducing the incidence of chronic diseases among middle-aged and elderly people by 18.4% (Table A1). However, when the main development direction of urbanization is the urbanization of the residential surroundings, healthcare has the greatest reducing effects on the health dangers induced by industrial sulfur dioxide (Table 5), with the incidence of chronic diseases decreased by 14.9% among the middle-aged and elderly adults (as shown by the marginal effect value of 0.149 in Table A1). For industrial wastewater, this mitigation effect is most significant only when the main development direction is the urbanization of living conditions, and the significance level is 1% (Table A1).

3.2.2. The Impact of City Type on Chronic Diseases

The imbalance of China’s urbanization levels leads to disequilibrium of the economic development level and the industrial structures, resulting in differing health impacts. This paper uses Nelson’s classification method from the United States [46] to classify 114 cities into three different types, based on data from 2011, 2013, and 2015. A city for which the proportion of the secondary industry in GDP is above the average level plus one standard deviation is classified as an industrial city, one for which the proportion of the tertiary industry to GDP is higher than the average level plus one standard deviation is classified as a commercial city, and the rest are classified as a mixed-economy city. This paper uses three groups of regression models to estimate results for the three types of cities, as reported in

Table 6. The health effects of urbanization in different city types.

Variable	Industrial City				Commercial city
	(1) PM2.5	(2) SO2	(3) Soot	(4) Wastewater	(1) PM2.5	(2) SO2
CU	0.044 ** (0.004 −0.514)	0.071 ** (0.008–0.646)	0.091 * (0.006–1.465)	0.183 (0.021–1.618)	0.310 (0.031–3.124)	0.244 (0.027–2.203)
Mhs	0.316 ** (0.128–0.781)	0.810 (0.321–2.043)	0.582 (0.245–1.383)	1.193 (0.480–2.964)	1.392 (0.574- 3.381)	1.236 (0.486–3.146)
ln (PM2.5)	1.373 *** (1.155–1.632)				1.037 (0.885- 1.214)
ln (SO2)		0.949 (0.889–1.014)				0.936 (0.860–1.019)
ln (Soot)			0.870 *** (0.800–0.946)
ln (Wastewater)				0.981 (0.861–1.116)
Mhs × ln (PM2.5)	0.016 *** (0.002–0.131)				2.380 * (0.969–5.848)
Mhs × ln (SO2)		0.663 (0.381–1.153)				0.669 (0.395–1.131)
Mhs × ln (Soot)			0.174 *** (0.074–0.409)
Mhs × ln (Wastewater)				1.437 (0.583–3.542)
Variable	Commercial city		Mixed-economy city
	(3) Soot	(4) Wastewater	(1) PM2.5	(2) SO2	(3) Soot	(4) Wastewater
CU	0.610 (0.059–6.331)	0.366 (0.042 −3.158)	0.064 *** (0.028–0.147)	0.057 *** (0.025–0.128)	0.122 *** (0.049 −0.303)	0.047 *** (0.021–0.106)
Mhs	1.388 (0.577–3.337)	1.317 (0.516 −3.360)	1.360 (0.891–2.075)	1.824 *** (1.224–2.718)	1.499 ** (1.007–2.232)	1.598 ** (1.076–2.372)
ln (PM2.5)			0.975 (0.923–1.030)
ln (SO2)				0.968 ** (0.944–0.993)
ln (Soot)	1.050 (0.951–1.158)				1.017 (0.989–1.046)
ln (Wastewater)		0.937 (0.822–1.069)				0.911 *** (0.877–0.947)
Mhs × ln (PM2.5)			2.250 ** (1.193–4.242)
Mhs × ln (SO2)				0.350 *** (0.254–0.482)
Mhs × ln (Soot)	0.786 (0.453–1.366)				0.176 *** (0.123–0.251)
Mhs × ln (Wastewater)		1.535 (0.879–2.680)				0.727 (0.482–1.097)

Notes: ***, **, and * represent the significance at 1%, 5%, and 10% levels, respectively; 95% confidence interval in parentheses; OR above parentheses.

. First, although improved healthcare fails to cancel the health risks caused by PM_2.5, the coefficient and marginal effect of the interaction term between healthcare and PM_2.5 are smallest in a commercial city (commercial city: OR = 2.380, 95% CI: 0.969–5.848; mixed-economy city: OR = 2.250, 95% CI: 1.193–4.242). In other words, at 13.2%, a commercial city has the lowest increase in the morbidity of chronic diseases among middle-aged and elderly people (

Table A2. The estimation of marginal effects on chronic diseases in the different types of city.

Different Types of City	Variable	Marginal Effects
		A Person Does Not Have Chronic Diseases	A Person Has 1 or 2 Chronic Diseases	A Person Has 3 or More Chronic Diseases
Industrial city	Mhs × ln (PM2.5)	0.845 *** (0.217)	−0.242 *** (0.066)	−0.602 *** (0.157)
	Mhs × ln (SO2)	0.082 (0.060)	−0.024 (0.017)	−0.058 (0.043)
	Mhs × ln (Soot)	0.382 *** (0.091)	−0.110 *** (0.028)	−0.272 *** (0.066)
	Mhs × ln (Wastewater)	−0.069 (0.095)	0.020 (0.027)	0.09 (0.068)
Commercial city	Mhs × ln (PM2.5)	−0.167 * (0.092)	0.035 * (0.020)	0.132 * (0.073)
	Mhs × ln (SO2)	0.082 (0.053)	−0.017 (0.011)	−0.065 (0.042)
	Mhs × ln (Soot)	0.034 (0.059)	−0.007 (0.012)	−0.027 (0.070)
	Mhs × ln (Wastewater)	−0.088 (0.056)	0.018 (0.012)	0.070 (0.045)
Mixed-economy city	Mhs × ln (PM2.5)	−0.195 ** (0.073)	0.039 ** (0.015)	0.156 ** (0.058)
	Mhs × ln (SO2)	0.256 *** (0.036)	−0.051 *** (0.008)	−0.205 *** (0.029)
	Mhs × ln (Soot)	0.354 *** (0.037)	−0.070 *** (0.008)	−0.284 *** (0.030)
	Mhs × ln (Wastewater)	0.103 * (0.046)	−0.021 * (0.009)	−0.082 * (0.037)

Notes: ***, **, and * represent the significance at 1%, 5%, and 10% levels, respectively; the parentheses are robust standard errors.

). Second, the mitigation effects of healthcare for the health problems induced by industrial sulfur dioxide and industrial soot and dust are greatest in a mixed-economy city (industrial soot and dust in an industrial city: OR = 0.174, 95% CI: 0.074–0.409; in a commercial city: OR = 0.786, 95% CI: 0.453–1.366; in a mixed-economy city: OR = 0.176, 95% CI: 0.123–0.251), which showed decreases in chronic diseases among the middle-aged and elderly people by 27.3% and 16.4%, respectively (Table A2). For the industrial wastewater index, the results reported in Table 6 and Table A2 show that the mitigation effect of healthcare on the health hazards brought about by industrial wastewater is insignificant. A possible explanation for this is that the treatment rate of industrial wastewater is constantly increasing with the rapid urbanization process. It is reported that, by 2016, the treatment rate of industrial wastewater in China had reached approximately 95%, so the improvement of health care conditions has little influence on the health problems caused by industrial wastewater.

3.2.3. The Regional Differences in the Impact of Urbanization on Chronic Diseases

Finally, this paper considers the spatial discrepancy of the urbanization level in China by separating the sample into three subsamples: Eastern, Central, and Western. In general, urbanization is highest in eastern China, followed by central and western China [47]. This classification method is frequently adopted and is useful for analyzing the regional differences in the impact of urbanization on health. The results shown in

Table 7. The effects of urbanization on chronic diseases in the different regions.

Variable	Eastern				Central
	(1) PM2.5	(2) SO2	(3) Soot	(4) Wastewater	(1) PM2.5	(2) SO2
CU	1.179 (0.163–8.546)	1.730 (0.245–12.208)	7.556 ** (1.036–55.126)	4.187 (0.582–30.098)	3.074 (0.513–18.430)	2.174 (0.369–12.793)
Mhs	1.198 (0.543–2.640)	1.083 (0.486–2.410)	0.897 (0.400–2.012)	0.661 (0.292–1.499)	0.287 *** (0.168–0.490)	0.565 ** (0.335–0.954)
ln (PM2.5)	1.162 *** (1.066–1.266)				1.000 (0.876–1.142)
ln (SO2)		0.980 (0.941–1.020)				0.923 *** (0.885–0.961)
ln (Soot)			1.157 *** (1.095–1.222)
ln (Wastewater)				0.981 (0.861–1.116)
Mhs × ln (PM2.5)	1.039 (0.423–2.552)				6.367 ** (1.540–26.320)
Mhs × ln (SO2)		0.582 *** (0.389–0.870)				0.637 ** (0.426–0.952)
Mhs × ln (Soot)			0.463 *** (0.300–0.717)
Mhs × ln (Wastewater)				2.531 *** (1.567–4.090)
Variable	Central		Western
	(3) Soot	(4) Wastewater	(1) PM2.5	(2) SO2	(3) Soot	(4) Wastewater
CU	2.826 (0.460–17.372)	2.070 (0.327–13.098)	2.173 (0.356–13.248)	1.401 (0.231–8.502)	23.288 *** (3.360–161.400)	0.712 (0.113–4.464)
Mhs	0.339 *** (0.204–0.564)	0.445 *** (0.266–0.744)	1.210 (0.626–2.341)	1.134 (0.591–2.174)	0.768 (0.394–1.498)	0.909 (0.461–1.792)
ln (PM2.5)			0.923 ** (0.857–0.994)
ln (SO2)				1.010 (0.972–1.051)
ln (Soot)	0.983 (0.936–1.032)				1.026 (0.986–1.067)
ln (Wastewater)		1.030 (0.959–1.107)				0.056 *** (0.022–0.143)
Mhs × ln (PM2.5)			1.210 (0.626–2.341)
Mh s× ln (SO2)				0.265 *** (0.173–0.407)
Mhs × ln (Soot)	0.316 *** (0.195–0.512)				0.197 *** (0.114–0.341)
Mhs × ln (Wastewater)		0.951 (0.443–2.042)				0.846 *** (0.791–0.906)

Notes: *** and ** represent the significance at 1% and 5% levels, respectively; the 95% confidence interval is in the parentheses; the OR is above the parentheses.

indicate that, in three regions, consolidated urbanization significantly increased the morbidity of chronic diseases among middle-aged and elderly people, but the extent of influence is most obvious in western China (CU: OR = 23.288, 95% CI: 3.360–161.400). Additionally, although the medical care does not offset the health risks caused by PM_2.5, the health risks are lowest in eastern China (Eastern: OR = 1.039, 95% CI: 0.423–2.552; Central: OR = 6.367, 95% CI: 1.540–26.320; Western: OR = 1.210, 95% CI: 0.626–2.341), which showed an increase of 0.5% in the incidence of chronic diseases among middle-aged and elderly people (

Table A3. The estimation of marginal effects on chronic diseases in the different regions.

Different Regions	Variable	Marginal Effects
		A Person Does Not Have Chronic Diseases	A Person Has 1 or 2 Chronic Diseases	A Person Has 3 or More Chronic Diseases
Eastern	Mhs × ln (PM2.5)	−0.008 (0.099)	0.003 (0.036)	0.005 (0.063)
	Mhs × ln (SO2)	0.117 ** (0.044)	−0.043 ** (0.016)	−0.074 ** (0.028)
	Mhs × ln (Soot)	0.166 *** (0.048)	−0.060 *** (0.018)	−0.106 *** (0.031)
	Mhs × ln (Wastewater)	−0.200 *** (0.053)	0.073 *** (0.019)	0.123 *** (0.034)
Central	Mhs × ln (PM2.5)	−0.359 * (0.140)	0.050 * (0.021)	0.310 * (0.121)
	Mhs × ln (SO2)	0.085 * (0.039)	−0.012 * (0.006)	−0.073 * (0.034)
	Mhs × ln (Soot)	0.215 *** (0.047)	−0.030 *** (0.008)	−0.186 *** (0.041)
	Mhs × ln (Wastewater)	0.008 (0.075)	−0.001 (0.010)	−0.007 (0.065)
Western	Mhs × ln (PM2.5)	−0.231 ** (0.072)	0.007 (0.006)	0.224 ** (0.070)
	Mhs × ln (SO2)	0.246 *** (0.040)	−0.007 (0.006)	−0.234 *** (0.039)
	Mhs × ln (Soot)	0.298 *** (0.051)	−0.009 (0.007)	−0.290 *** (0.050)
	Mhs × ln (Wastewater)	0.533 *** (0.088)	−0.017 (0.012)	−0.516 *** (0.085)

Notes: ***, **, and * represent the significance at 1%, 5%, and 10% levels, respectively; the values in parentheses are robust standard errors.

). In contrast to PM_2.5, healthcare is the most effective in health hazards caused by industrial soot and dust in eastern China (Eastern: OR = 0.463, 95% CI: 0.300–0.717; Central: OR = 0.316, 95% CI: 0.195–0.512; Western: OR = 0.197, 95% CI: 0.114–0.341), where the morbidity of chronic diseases among middle-aged and elderly adults decreased by 10.6% (Table A3). Of the three regions, the central region of China had the greatest medical mitigation effect on the health problems caused by industrial sulfur dioxide (Eastern: OR = 0.582, 95% CI: 0.389–0.870; Central: OR = 0.637, 95% CI: 0.426–0.952; Western: OR = 0.265, 95% CI: 0.173–0.407), with the incidence of chronic diseases decreased by 7.3% (Table A3). More interestingly, the enhancement of medical care most effectively reduced the health dangers caused by industrial wastewater in western China, where the prevalence of chronic diseases among the elderly population was reduced by 51.6% (Table A3).

3.3. The Robustness Test

To maintain the robustness of the results, it is important to consider the robustness of the results from various aspects. Therefore, this paper focuses on both the substitution dependent variable and the endogeneity; on the one hand, we account for the multidimensionality of health in middle-aged and elderly people, and on the other hand, personal avoidance behaviors are considered. The part of the robustness test mainly explores these two aspects. The first endogenous question is the residential health measure. It uses the objective health measure self-rated health as the dependent variable and considers the impact of the interaction term between medical care service and environmental pollution on this measure among elderly people. The results show in

Table 8. The influence of urbanization on self-rated health.

Variable	Model 1	Model 2	Model 3	Model 4
CU	−4.166 *** (0.366)	−4.571 *** (0.363)	−4.161 *** (0.391)	−4.673 *** (0.364)
Mhs	0.757 *** (0.167)	0.582 *** (0.161)	0.566 *** (0.161)	0.485 *** (0.162)
ln (PM2.5)	−0.178 *** (0.026)
ln (SO2)		−0.006 (0.013)
ln (Soot)			−0.004 (0.013)
ln (Wastewater)				−0.065 *** (0.018)
Mhs × ln (PM2.5)	−0.325 (0.260)
Mhs × ln (SO2)		−0.456 *** (0.126)
Mhs × ln (Soot)			−0.766 *** (0.121)
Mhs × ln (Wastewater)				0.266 * (0.153)

Notes: ***and * represent the significance at 1% and 10% levels, respectively; the parentheses are robust standard errors; the control variables are the same as those in Table 1, but only the results for the control variables are reported due to space limitations.

estimated results that are generally consistent with the baseline model; that is, the improvement of the healthcare conditions can significantly reduce the negative effects of environmental pollution (sulfur dioxide, soot and dust) on self-rated health among middle-aged and elderly people. The only exception is that for, the environmental pollution index PM_2.5, the impact of medical care and air pollution on the residential self-rated health is different from the baseline model. This could be due to sampling error or statistical error, or it is possible that respondents may incorrectly assess their health.

The second endogenous issue considered is individual avoidance behavior. Specifically, residents may move to areas with lower air pollution in order to avoid individual health damage. In order to consider population mobility, this paper further restricted the sample by excluding respondents who had lived outside of their permanent residence for more than 6 months. Compared to the baseline regression, the sign and significance of coefficients are roughly insignificant, but the magnitude of the interaction term coefficients is higher (

Table 9. The influence of urbanization on chronic diseases after control of population mobility.

Variable	Model 1	Model 2	Model 3	Model 4
CU	−2.166 *** (0.390)	−2.660 *** (0.389)	−1.677 *** (0.427)	−2.573 *** (0.390)
Mhs	0.238 (0.173)	0.466 *** (0.170)	0.275 (0.169)	0.163 (0.169)
ln (PM2.5)	−0.010 (0.027)
ln (SO2)		−0.044 *** (0.012)
ln (Soot)			0.010 (0.013)
ln (Wastewater)				−0.101 *** (0.018)
Mhs × ln (PM2.5)	0.936 *** (0.252)
Mhs × ln (SO2)		−1.047 *** (0.123)
Mhs × ln (Soot)			−1.152 *** (0.122)
Mhs × ln (Wastewater)				0.126 (0.156)

Notes: *** represent the significance at 1% levels; the values in parentheses are robust standard errors; the control variables are the same as those in Table 1, but only the results for the control variables are reported due to space limitations.

), which suggests that the mitigation effects of healthcare service levels on the health risks induced by air pollution are likely to be underestimated without accounting for population mobility.

4. Conclusions

This paper employs the fully arrayed polygonal graphical indication method to construct comprehensive urbanization indicators (based on the four dimensions of demographic urbanization, economy urbanization, residential surroundings urbanization, and residential conditions urbanization) and a medical treatment index. Then, we incorporate urbanization indicators, environmental pollution indicators, healthcare, and the interaction terms of environmental pollution and healthcare into the ordered-logistics model to explore the magnitude of the impacts of urbanization on health, including the negative impact of environmental pollution and the positive impact of improved medical care. We also study the heterogeneity effects based on multidimensional urbanization, regional differences, and different city types. Based on this analysis, we conclude the following:

(1)

The comprehensive urbanization index was constructed by applying the fully arrange polygon graphical index method from four dimensions: the urbanization of the population, the urbanization of economy, the urbanization of residential environment, and the urbanization of residential conditions. The results indicate that integrated urbanization can significantly decrease the rate of chronic diseases among the middle-aged and elderly. However, although the enhancement of healthcare conditions can significantly reduce the prevalence of chronic diseases induced by industrial sulfur dioxide and industrial soot and dust, it cannot offset the health damage caused by PM_2.5.

(2)

Each city has different development directions in the process of urbanization, which can be considered by the four dimensions of demographic urbanization, economy urbanization, residential surroundings urbanization, and residential conditions urbanization. When the main development direction is economic urbanization, although the medical care still does not fully counteract the health risks from PM_2.5, the health damage is the lowest, with an increase in the morbidity of chronic diseases among middle-aged and elderly adults of only 12.4%. In addition, the mitigation effect of healthcare for health hazards caused by industrial soot and dust is the highest, with the largest decrease in the incidence of chronic diseases at 18.4% in the primary stage of urbanization. When the development direction is residential surroundings urbanization, the healthcare service has the greatest mitigating effect on the health problems caused by industrial sulfur and dioxide, reducing the incidence of chronic diseases in middle-aged and elderly people by 14.9%. With residential condition urbanization, healthcare can significantly decrease the incidence of chronic diseases caused by industrial wastewater, and the significance level is 1%.

(3)

Referring to Nelson’s classification in the United States, the study cities are separated into three categories: industrial, commercial, and mixed-economy cities. The results suggest that although medical treatment cannot counteract the health risks induced by PM_2.5, at 13.2%, the commercial cities have the lowest overall increase in the incidence of chronic diseases among middle-aged and elderly people. The greatest reduction in the incidence of chronic diseases in middle-aged and elderly people induced by industrial sulfur dioxide and industrial soot and dust is seen in mixed-economy cities, with incidences decreased by 27.3% and 16.4%, respectively.

(4)

In order to consider the regional imbalance of urban development in China, the sample is divided into three sections: eastern, central, and western China. The regression results suggest that in eastern China, although healthcare does not offset the health damage caused by PM_2.5, the overall health risks are lowest, with the incidence of chronic diseases increase by 0.5%. The mitigation of health hazards caused by industrial sulfur and dioxide is the lowest in eastern China, with the prevalence of chronic diseases decreased by 7.4%. However, the reduction influence of medical treatment on health risks caused by industrial soot and dust is the largest in the central part of China, which shows an 18.6% decrease in the prevalence of chronic diseases. Healthcare can significantly decrease the incidence of chronic diseases induced by industrial wastewater.

In light of these conclusions, this paper argues that local governments should focus on controlling air pollutants (particularly PM_2.5) and industrial wastewater pollution. First, with regard to PM_2.5, local governments can compensate for the health hazards of PM_2.5 by enhancing the economic development level and narrowing the development levels between regions. Furthermore, the government should take into account improving the living environment and economic level, balancing the development of secondary and tertiary industries, and strengthening the configuration of medical facilities to tackle the industrial sulfur dioxide and industrial soot and dust pollution. Finally, when treating industrial wastewater pollution, the local government can take some measures to improve the living environment and living conditions, balance the development of industries, and improve other measures to focus on the treatment of eastern and central China and cities dominated by the secondary and tertiary industries.

Although this study explores the dual impact that urbanization has on health through improved healthcare and increased environmental pollution by incorporating healthcare services, environmental pollution, urbanization, and health into the same framework, the data studied are cross-sectional and do not account for changes in dynamics. In future studies, other types of data may be considered to more precisely examine the relationship between the dynamics of urbanization and health.