The Impact of Environmental Pollution on the Health of Middle-Aged and Older Adults in China

While several studies have demonstrated the negative impacts of environmental pollution on population health, in general, few studies have examined the potential differential effects on the health of middle-aged and older populations, i.e. 45 years and older. Given the twin concerns of environmental pollution and population aging in China, this article employed a xed effects model to infer the impact of environmental pollution on public health with a particular focus on middle-aged and older adults. The analyses were based on data from the 2011–2018 waves of the CHARLS and pollutant data from prefecture-level cities. The results showed that environmental pollution signicantly increased the risk of chronic diseases and negatively impacted the health of middle-aged and older adults. Environmental pollution had its greatest negative effect on the health of the elderly, women, urban residents and those with lower incomes than for their counterparts. We further found that the main channels of effect were through reduced physical exercise and an increase in depressive symptoms, and the pollution prevention actions alleviated the health deterioration of environmental pollution for the middle-aged and elderly. It is imperative for the government to urgently reinforce policy's enforcement to decrease air and water pollution, and enhance the ability to circumvent pollution for the lower socioeconomic groups.


Introduction
The growing prevalence of chronic conditions over the last 20 years has become a serious health problem and the main cause of premature mortality (Zhong and Xiang, 2010; Li and Ping, 2016). It is widely recognized that non-communicable chronic diseases have become more prevalent than acute infectious diseases in China and in the rest of the world. Estimates suggest that chronic diseases in China accounted for 86.6% of deaths and 70% of the total burden of disease (Xiong, 2019). Chronic illnesses not only seriously affect the quality of life of individuals and their families, but also have a huge impact on health systems and the economy. Health systems around the world face a daunting task in managing the quality of health service provision for chronic conditions and their long-term consequences (Epping-Jordan et al., 2004).
An extensive review of the literature reveals that the prevalence of chronic conditions is associated with lower social and economic status (Xia and Li, 2018;NoorUlHuda, 2020). For example, studies from China have shown a link between urbanization and the prevalence of diabetes (Attard et al., 2012;. Similarly, the development of chronic illnesses is closely correlated with the increasing prevalence of obesity and unhealthy behaviors (e.g., tobacco use, poor nutrition, and the lack of physical activity) and worsening environmental conditions (Siavash et  Rapid industrial growth has led to an increase in waste emissions in China. The range of health effects stemming from environmental pollution have been well documented. Air pollution has been shown to be associated with the birth of underweight babies (Gong et al., 2018), a growth in infant mortality (Lelieveld et al., 2015;Nihit et al., 2019;Fotourehchi, 2016), and a greater incidence of otitis media infections in early childhood (Deng et al., 2017). Studies have also found that adults' health has been impacted. For instance, industrial air pollution has been shown to elevate mortality rates, shorten life expectancy, and reduce disability-adjusted life years (Ahmad et al., 2020;Pope et al., 2013;Peng et al., 2020;Huang et al., 2018). These effects have tended to result in increased hospitalizations and higher health expenditures (Cui et al., 2016;Zeng and He, 2016). Water pollution has been linked to water-borne diseases and chemical poisoning (Han and Yan, 2014), and negatively associated with mental and self-reported health outcomes (Wang and Yang, 2016). However, there is a paucity of studies on the impact of environmental pollution on the prevalence of chronic diseases, in general, rather than on speci c chronic diseases, such as asthma and other related respiratory problems in children (Sram et Wang and Yang, 2016). We limit our attention to middle-aged and older adults in China, since they are more susceptible to environmental pollution and are likely to suffer from multiple chronic diseases (Freedman and Martin, 2000;Wolff et al., 2002). It is hoped that the insights from our study will inform policy decision making on a range of pollution measures and mechanisms and thereby have a positive impact on the health of Chinese residents.
To shed light on this important issue, we highlight the general effects of environmental pollution on chronic diseases using data drawn from the China Health and Retirement Longitudinal Survey (CHARLS).
The potential differential effects of environmental pollution on the health of different age groups are assessed. We also examine the channels through which the prevalence of chronic conditions are indirectly affected by environmental pollution using a mediation model. Finally, we consider whether pollution control policies alleviate the negative consequences of environmental pollution using a moderation effects model.
The study is organized as follows. The next section introduces the data and the empirical strategy employed. The empirical results are presented in Sect. 3. Section 4 discusses the ndings in the context of the literature and identi es policy implications, and Sect. 5 offers a brief set of conclusions.

Data
Information on the health status of individuals as well as their age-sex and family structure were derived from the China Health and Retirement Longitudinal Survey (CHARLS). This is a longitudinal survey based on nationally representative samples of Chinese residents aged 45 and over. Four waves of this survey have been conducted from 2011 to 2018 through use of a multistage random cluster design (Shan et al. 2016;Zhao et al., 2014). The CHARLS contains information on 17,500 individuals and 10,000 households in 150 counties/districts and 450 villages/resident committees in 28 provinces and municipal cities.
These areas vary substantially in geography, economic development, public resources, and health indicators. By 2018, the 28 sample units included in the CHARLS constituted 98.59% of China's population. The survey provides detailed information on respondents' health, demographics, socioeconomic status, and other related information both at the individual and household level. It presents a rich array of information on chronic illnesses that makes it ideal for analyzing the effect of environmental pollution on the physical health of middle-aged and older people.

Measures
We examined the effect of environmental pollution on health, similar to the previous literature, but with a special focus on those aged 45 years and older. For the dependent variable, we relied on information on chronic illnesses documented in the CHARLS to de ne the health status of middle-aged and older people.
The information derives from physical examinations or diagnoses by doctors (Shan et al., 2016). Thus, the diagnosed chronic disease data represents an objective measure of ill health and therefore may be free of "justi cation bias" (Ning et al., 2016;Fan et al., 2019). We created a continuous variable to represent the "number of chronic diseases," including hypertension, dyslipidemia, diabetes or high blood sugar, malignant tumor, chronic lung diseases, liver disease, heart attack, stroke, kidney disease, and asthma. . We did not consider inhalable PM such as PM2.5 and PM10 specially, because data availability was restricted to the prefecture-level. We evaluated SO 2 , PM, and water pollution emissions measured in per unit area (ten thousand tons per square kilometers) in each prefecture to map the prefecture-level cities' precise environmental pollution level (Tao et al., 2019).
We also controlled for a series of individual-level, household-level, and other environmental characteristics. The individual-related variables included age, gender, marital status, educational attainment, health behaviors (smoking, alcohol consumption, and social activities), medical insurance coverage, and individual income. Household-related variables included the residential area and whether the respondent was living with their children. We also controlled for the prefecture level annual average air temperature, because this can also affect health status (Cui et al., 2016).

Statistical Analyses
We examined the effects of air and water pollution on chronic disease status for adults aged 45 years and older using multiple Poisson regression analysis. The number of chronic diseases was measured on count data. Poisson regression has the advantage of tting nonlinear models over the linear regression models, because it deals with situations in which the dependent variable is a count (Fagbamigbe and Adebowale, 2014). Additionally, there were large individual variations in ill health, so we used a xedeffect model to control for time-invariant unobserved factors in the unbalanced panel data. We also used standard errors clustered at the individual level to account for the multilevel, multistage sampling issues, and controlled for potential individual heteroscedasticity to obtain consistent parameter estimates (Cui et al.,2016;Stock and Watson,2008). The xed-effect Poisson regression model was speci ed as follows: where the outcome, ill-health it , stands for the number of chronic diseases for individual i at time t. The key explanatory variable Pollutant it indicates the air and water pollutants, that is, the SO 2 and PM, and wastewater emissions per unit area of each prefecture. X it is a vector of individual and household observable characteristics and environmental variables, such as age, gender, educational levels, health behaviors, individual income, cohabitants, annual temperature. The term µ i represents individual-related xed-effects accounting for all time-invariant factors that may affect health, and ε it is a random error term. Because the Poisson regression is a nonlinear model, we calculated the marginal effects of environmental pollution on individuals' ill health to obtain more intuitive and practical results. Table 1 reports the descriptive statistics used in the analysis and de nes each variable. The average number of chronic diseases was 0.684. The annual average industrial emissions were 74.67 thousand tons of SO 2 /km 2 , 27.85 thousand tons of PM/km 2 , and 7.74 thousand tons of wastewater/km 2 , respectively. More than half of the respondents were between 45 and 60 years of age, and most had a primary school level of education. Most (86.5%) were married, and around 60% lived in rural areas. Full analysis Table 2   Note: *** P < 0.01, ** P < 0.05, * P < 0.1. Clustered-robust standard errors are in parentheses.

Statistical description
Model 2 ( Table 2) presents estimates of a robustness test, to assess how sensitive the ndings were to alternative measure of ill health. Speci cally, we de ned a dichotomous variable presence or absence of chronic disease, which was coded as 1 if a respondent had been diagnosed with any chronic disease in the last year otherwise zero. This variable was used to examine the robustness of the health effects of environmental pollution. The marginal effects from the xed-effect model are showed in Columns 4−6 of Table 2. The probability of having any diagnosed chronic disease increased 0.16%, 0.78%, and 1.56% with higher emissions of SO 2 , PM, and wastewater, respectively. These ndings con rmed the robustness of the central conclusion.

Subsample analysis
We further explored the extent to which the effects of environmental pollutants on the number of chronic diseases varied within different subgroups. We used the xed-effect Poisson model, and presented the marginal effects (Fig. 1). We examined the health effects by individual-level characteristics, that is, by age and sex. In particular, we analyzed 3 age groups: 45 to 59 years old, 60 to 74 years old, and 75 years and above. These re ected the fact that the body and immunity status may differ for those who are middleaged, those who are elderly, and nally, those who are even older (Zhong and Xiang, 2010). The negative effects of environmental pollution on health appeared to be greater among the elderly between 60 to 74 (see Fig. 1A). This may be attributed to the fact that middle-aged adults have higher levels of protective immunity to, while the vey elderly are more likely to stay indoors (and so are less likely to be exposed to environmental pollution) (Lü et al., 2015; Huang et al 2020). Also, the results in Fig. 1B show a larger negative impact of air and water pollution on health among women. We also investigated the differential effects of household income and residential area. We classi ed the sample into low and high-income groups using the 50th percentile of household income. We found that SO 2 , PM, and wastewater emissions per unit area signi cantly increased the number of chronic diseases by 0.60%, 2.67%, and 15.51%, respectively in the low-income group, but the effects were insigni cant for those in the highincome group (see Fig. 1C). This may be due to the fact that poor individuals are more likely to be exposed to pollution as they have no means of staying away from environmental pollution, for instance by migrating or purchasing air puri cation equipment. (Lü et al., 2015). Additionally, air and water pollutants have a stronger negative impact on people living in urban areas (see Fig. 1D). This may be attributed to the fact that pollution is worse in urban areas than it is elsewhere (Li et al., 2020).

Mechanism analysis
In this section, we reveal the mechanisms behind the negative effects of environmental pollution on health. According to previous studies, environmental pollution can pose a barrier to involvement in outdoor physical activities (Lü et  depressed in answers to each question, and 0 otherwise. A summed score of the responses to these ten items was used to indicate depression, with higher scores meaning a greater degree of depression. We estimated the mechanisms through the use of a mediation model. In particular, we rst calculated the effect of environmental pollutants on the mediators for "physical exercise" and "depressive symptoms" (see Panel A in Table 3). Models 1 and 2 in Panel A revealed that air and water pollutants emission per unit area signi cantly reduced the probability of physical exercise participation and increased the CESD scores. Consequently, we used physical exercise and depression symptoms as the mediator variables in the next step.   Table 2. Note: *** P < 0.01, ** P < 0.05, * P < 0.1. Clustered-robust standard errors are in parentheses. The other control variables were the same as shown in Table 2. Table 3

Moderation analysis
The adverse impact of air and water pollution on individuals' health has become an increasingly serious problem. The Chinese government has repeatedly stressed the importance of environmental protection, which may alleviate the negative effect of exposure to industrial pollution on citizens' physical health. A moderator is a variable that alters the strength or direction of the relationship between the environmental pollution and health (Holmbeck, 1997). We discovered that air and water pollutants emissions fell after 2013, which can be explained by the fact that the Chinese central government under Mr Xi took greater efforts to reduce pollution (Gao et al., 2016). We therefore used the pollution control policies issued in 2013 as the moderator. Because environmental improvement policies become more in uential over time, we created the variable "policy" coded as the number of years since these policies were implemented in 2011, 2013, 2015, and 2018, which ranged from 0 to 6. We added the interaction term to the health equation, which was created by multiplying the environmental pollutants emissions and the moderator (Gao et al., 2016;Kamimura et al., 2017). We further examined the role of pollution control policies with the moderation model.
We used the interaction term created by multiplying the SO 2 , PM and wastewater emissions and the moderator separately (Holmbeck, 1997;Kenny and Judd, 1984), to examine the role of pollution control policies in reducing the negative effect of environmental pollution. The results in Model 3 of Table 4 show that the negative effect of SO 2 , PM, and wastewater reduced by 0.17%, 0.38%, and 0.69%, respectively, through the pollution control policies indicator. Therefore, the pollution control policies did signi cantly reduce the negative relationship between environmental pollution and chronic diseases among middleaged and older people. Note: *** P < 0.01, ** P < 0.05, * P < 0.1. Clustered-robust standard errors are in parentheses. The other control variables were the same as shown in Table 2.

Discussion
The present study considered how the air and water pollution affected the health of adults aged 45 years and older in China. We found clear evidence that environmental pollution was associated with poorer health status. In particular, air and water pollutants signi cantly increased both the risk of chronic diseases as well as the number of chronic diseases. These ndings were consistent with other studies. We discovered that the effects of environmental pollution varied according to age, sex, household income, and residential area. In particular, the greatest negative effect of air and water pollution on health was found among 60−74-year-olds, females, adults of lower income, and urban residents. Furthermore, we revealed two possible channels through which the air and water pollutants reduced middle-aged and old people's physical activities and increased their depression levels, ultimately leading to worse chronic conditions. Finally, we found pollution control policies did alleviate the negative impact of environmental pollution on the health of the middle-aged and the elderly.
Our ndings have a number of implications for public policies designed to reduce environmental pollution and its health costs and to improve the health of the elderly. The moderator effect of pollution control policies suggests the need for developing, testing, and implementing an appropriate mix of policy instruments to prevent, control, or dramatically reduce environmental pollution, as well as the for monitoring the e cacy of air and water pollution prevention efforts. For example, investment to increase green space is essential. This can bene t residents' psychological well-being, encourage middle-aged and old people to undertake more physical activity, and reduce air pollution (Gascon et al., 2016;Liu et al., 2017). In addition, adults of lower income can be plunged into poverty and poorer health because of environmental pollution. A vicious circle is thus created. A comprehensive plan for environmental protection, health reform, and poverty alleviation programs should be tailored to the low-income elderly to dissipate the negative effect of environmental hazards on health.
The present study has several limitations. First, because of a lack of operational data at the prefecture level, we were unable to use other air pollution indicators, such as respirable PM (PM2.5 and PM10). We were also unable to establish concentrations of air pollution. Second, we were limited to industrial air and water pollutants again because of a lack of (access to) data. Future studies could evaluate the effect of other environmental pollutants to which people are exposed in their daily lives, for instance localized, tra c, and domestic pollutants. Third, it would be helpful to identify the spatial spillover effects of pollution, because pollution is apt to diffuse and migrate across different regions. In particular, it would be valuable to do this at provincial rather than prefecture level. The spatial spillover effects are not apparent in our study because most of the counties/districts surveyed in CHARLS are not adjacent to each other. Finally, we focused on physical chronic diseases because they are the most common cause of death among elderly Chinese people. Future researchers could examine in detail the subjective health damage caused to the elderly by environmental pollution.

Conclusions
This is the rst study to assess the effects of air and water pollution at the prefecture level on the chronic disease risk of the middle-aged and elderly in China. It has found quite robust evidence that SO 2 , PM, and wastewater emissions per unit area signi cantly increased the number of chronic diseases, indicating that they have a corrosive effect on health status. Among all the subgroups, the elderly aged between 60−74, females, adults of lower income, and urban residents faced the greatest risk of chronic disease from environmental pollution. The study further disentangled the channels of physical activity and symptoms of depression through which health is comprised by pollution. Finally, pollution control policies did reduce the negative impact of environmental pollution on the health status of the middleaged and elderly in China.

Declarations
Ethics approval and consent to participate: Not applicable.
Consent for publication: Not applicable.
Availability of data and materials: The datasets used during the current study are available from the corresponding author on reasonable request.
Con ict of interest: The authors declare that they have no competing interests.  Figure 1