The influence of environmental awareness and conditions on successful aging: Evidence of air and water pollution in China

ABSTRACT

This study examines three aspects of successful aging including subjective health, cognitive ability, and social participation and their relations with environmental conditions and environmental awareness. We used the seemingly uncorrelated regression and multilevel models to estimate the joint relationship by combining the Chinese General Social Survey (2013 and 2021) (N = 5404 and 1580 accordingly) and the Chinese Statistical Yearbook and used propensity score analysis to identify potential endogenous issues. The first finding is that older people who were men, married, lived in urban areas, or received social security had an edge in aging successfully. Secondly, the adverse effects of environmental pollution on older adults’ cognitive ability and social participation can be reduced by the protective effects of environmental awareness. Additionally, air pollution had more apparent effects, and water pollution was more modest in social involvement. This study offers empirical results to enhance the well-being of the older population by lifting their environmental perception and improving environmental quality.

KEYWORDS:

• Successful aging
• environmental awareness
• air pollution
• water pollution

Introduction

Population aging is a global public health concern (Ng et al., 2022). In 2017, the World Health Organization (WHO) estimated that over 700 million people worldwide were 65 or older, making up 9 percent of the world’s population, and would increase to around 1.5 billion by 2050, accounting for 16 percent of the world’s population (WHO, 2019).

Various terms and theories including successful aging (Rowe & Kahn, 1987), active aging (Kalache & Gatti, 2003), healthy aging (WHO, 1990), and productive aging (Kaye et al., 2003), have been developed to define the quality and state of the aging process. In essence, these theories seek to prescribe what older people should to achieve better aging outcomes do as they age. Among them, successful aging is acknowledged as the most frequently applied worldwide. The original and expanded model of successful aging includes three components: avoidance of disease or disabilities, maintenance of cognition, and active social involvement, which is the refinement and integration of the core content of the other three concepts (Rowe & Kahn, 1997). The latest evolvement of successful aging (SA) considers micro-mezzo-macro factors including biological factors, behavioural factors, social factors, and environmental factors that affect the quality of life of older adults.

Previous studies focus on individual-level predictors of SA, such as social demographic variables (e.g. education, SES) (Hodge et al., 2013), health beliefs and behaviours (Chodzko-Zajko et al., 2009) and social connections (Gallardo-Peralta et al., 2022). Differences due to gender advantages (Newman et al., 2003), marital status (Yalcinoz Baysal et al., 2020), and living areas (Hsu et al., 2010) are also examined. However, there is only limited research focusing on the health impact of environmental contamination among older Chinese (Di Ciaula & Portincasa, 2020).

Environmental pollution can affect health in various ways. In the short term, environmental pollution damages the human respiratory system (Buffart et al., 2018) and causes digestive system disorders (Gu et al., 2020), chronic bronchitis (Hooper et al., 2018), and emphysema (Hu et al., 2015). Long-term effects of exposure to a highly polluted environment include dementia (Paul et al., 2019), abnormalities of the nervous system (Costa et al., 2020), a significantly elevated risk of cancer (Rodrigues et al., 2021), and disability (Weuve et al., 2016).

However, the theoretical link between aging and environmental issues tends to overlook the particularity of the older population as integrated social actors rather than just their health. Especially for the study of China’s aging, the research bias is even more evident. China had a total population of 1.41 billion in 2020, and 190 million, or 13.5 percent of the population, were 65 years or older. China’s aging population has accelerated over the previous ten years, and the percentage of those 65 and older has climbed by 4.7 percentage points since the sixth population census in 2010 (Hu et al., 2020). The older population is a vulnerable group to pollution because of their frailty, yet, few studies have focused on the mechanisms of physical and social functions affected by pollution in older age groups.

Existing literature on the associated factors of SA in China is limited to socio-demographic and psychosocial resource factors (Hsu, 2007). Examining the environmental factors, this paper contributes to environmental and aging studies in three ways. First, the paper studies the internal mechanisms underlying the relationship between the mix of pollutants, environmental awareness, and SA by considering them within the same analytical framework, rather than investigating them separately and dealing with the endogeneity issue. This paper explores the relationship between macro and micro using integrated data, highlighting how major effects and moderating effects simultaneously influence the SA process. Second, this paper shows nuanced findings from a multidimensional perspective, including time-varying effects of pollution conditions and SA when comparing 2013 and 2021, and structural differences in pollution sources when comparing air and water pollution. Third, the paper broadens the subject of previous research by concentrating on how much environmental factors affect aging and how to improve elderly well-being in China, where the aging challenge is becoming increasingly prominent, by tying together ecological and social support policies.

Literature review

Three dimensions of successful aging

Physiological function

The component entails physical functioning impairment and overall health status (Annele et al., 2019; Boyes, 2013). Activities of daily living (ADL), instrumental activities of daily living (IADL), and chronic disease are frequently used to measure the former. Self-rated health, relative health perception, and life satisfaction are typically used to gauge the latter. Yamada and colleagues (2022) evaluated the self-rated health indicator to show the prevalence of SA in latent subgroups of community-dwelling individuals. The previous research is typically connected to Healthy Aging studies, which stress the well-being and lifespan of older people and seek to address their welfare needs by offering medically sound services (Campisi et al., 2019).

Cognitive function

It speaks of people’s perception, focus, memory, reasoning, language, interpersonal interactions, and personalities (Hartley et al., 2018; Wang et al., 2020). Muhammad and colleagues (2022) used the Mini-Mental State Examination to create five domains that encompass memory and arithmetic functions, and they discovered urban people’s advantages in cognitions. Midlife cognitive dysfunction is found to be both a crucial predictor and an indicator of aging in prior studies. Nevertheless, its comparison with the other dimensions has not been fully established. This paper considers language ability as an important feature of cognitive function.

Social function

Regular physical activity, outdoor activities, mobility performance, social engagement, and perceived social support are used to quantify this component (Lee et al., 2022), which is adjacent to Active Aging that emphasises the value of social participation among older people and pays attention to the supporting roles of family-friend interactions, community engagement, cultural and sports events, and public welfare activities (Mejía et al., 2017). However, a combined perspective of social networks and pathological nursing, along with questions concerning how to define and quantify social function, remains controversial.

The environment as a determinant of successful aging

Pollution

A healthy environment would promote longevity and slow the rate of ability decline, while an unhealthy environment may cause physical and mental harm. The relationship between pollution and health has been extensively discussed in epidemiology and pathophysiology (Schmidt, 2019). Air and water pollution are two primary signs of contamination. The fragility and vulnerability of older people are impacted by air pollution. It is found that exposure to PM2.5 is linked to an increased likelihood of fragility, emergency room visits, hospitalisation for coronary heart disease, and a decline in lung function (Amsalu et al., 2019). Hu and colleagues (2020) used a longitudinal analysis to discover a strong correlation between air pollution exposure and the frailty of older Chinese people. Applying China’s provincial panel data, Zhang and colleagues (2023) constructed spatial econometric models and found the increase in air pollution had a significant detrimental effect on average life expectancy. Chen and colleagues (2013) also provided a quasi-experimental approach to illustrate a causal relationship that the total suspended particulates in northern China caused reductions in life expectancies of 5.52 years. Water pollution impairs the quality of drinking water and causes food contamination by introducing dangerous toxic substances into foods and beverages (Chang et al., 2019). Awoke and the working group (2016) gathered ecological data from significant river basins in developing countries and discovered that urban water deterioration negatively impacted resident health. In a nation where water resources are inadequate and unevenly distributed, the Chinese population is compelled to use contaminated sources because of water scarcity, and the exacerbating industrial and municipal pollution helps to explain why the rate of digestive cancer is rising (Zhang et al., 2010). However, the pertinent research is restricted to discussing how air and water pollution jointly affect older people’s health, cognition, and social activities.

Pollution awareness

We can speculate that environmental awareness has a conditioning influence on human development drawn from the paradigm of neo-ecology and constructivism (McDowell, 2014). Environmental concern is a holistic process involving knowledge, comprehension, willingness, and aspiration of environmental issues, which eventually contribute to environmentally sustainable development (Tran, 2006). According to the perception-behavior model, people are more inclined to participate in public environmental governance and protection as they enhance their awareness of environmental issues. The pollution awareness concept complements the actual environmental situation by focusing on participants’ subjective impressions of environmental contamination.

Research questions and hypotheses

Current research is limited in the following two aspects: (1) The current studies investigate the population impact on one particular dimension of SA (usually a physical function), and leave out other important dimensions unexamined. (2) Evidence about the effects of water and air pollution on SA is scarce, and fewer studies examine the impact of environmental awareness on SA. The present study aims to address two research questions: (1) Do environmental factors affect SA? (2) Can pollution awareness moderate the effects of pollution on SA?

Guided by the theories of successful aging theory and environmental awareness, we propose the following hypotheses:

Hypothesis 1: Air pollution awareness is positively associated with successful aging. People with higher awareness of air pollution would have higher levels of subjective health, cognitive ability, and social participation.

Hypothesis 2: Water pollution awareness is positively associated with successful aging. People aware of water pollution have higher levels of subjective health, cognitive ability, and social participation.

The following hypotheses are about the interaction between pollution situation and pollution awareness:

Hypothesis 3: Individuals’ awareness of air pollution affects successful aging differently. In areas with more air pollution, the impact of pollution awareness on SA would be more salient.

Hypothesis 4: Individuals’ awareness of water pollution affects successful aging differently. In areas with more water pollution, the impact of pollution awareness on SA would be more salient.

Figure 1 demonstrates the conceptual framework to understand the dynamic relationship between the external environment and the internal perceptions of SA.

Figure 1. The conceptual framework for studying how environments affect successful aging.

Method

Sample

Data for this study were drawn from the Chinese General Social Survey (CGSS) in the years 2013 and 2021. CGSS is a national, synthetical, and consecutive survey in China that gathers information on people aged 18–65 years old regarding education, health, social attitudes, etc. to observe the characteristics and trends of the local society and discuss social issues with theoretical and practical significance. The CGSS employed a multi-stage stratified sampling procedure, with counties serving as primary sampling units, urban communities and rural villages as secondary sampling units, and households being randomly selected using a mapping sampling method (Bian & Li, 2012). We chose the years 2013 and 2021 because they contain measurements of environmental awareness, attitudes, and actions, and the time period offers a comparable view of the developing patterns in the environment. In the 2013 survey, all samples (N = 11,438) were asked environment-related questions, while in the 2021 survey, about 33.6 percent of respondents (N = 8,148) were randomly assigned to answer the environment part. The study focused on adults aged 50 and above, so we kept observations with the targeted age, resulting in a final sample of 6,984 (N = 5,404 in CGSS 2013, N = 1,580 in CGSS 2021). We also included data on provincial environmental conditions from the China Statistical Yearbook compiled by the National Bureau of Statistics of China. We used the 2014 and 2022 Statistical Yearbook to match 2013 and 2021 individual data accordingly since it provides extensive information on the evolution of national society during the previous natural year.

Measures

Successful aging

The measurement of SA contains three dimensions. The first is subjective health, measured by how people think of their current physical condition on a 5-point Likert-type scale from ‘very unhealthy ( = 1)’, ‘somewhat unhealthy ( = 2)’, ‘moderate ( = 3)’, ‘somewhat healthy ( = 4)’, to ‘very healthy ( = 5)’. The second is cognitive ability, which usually includes logical, mathematical, and language abilities (Lubinski, 2009). Language ability is regarded as the core function of cognition and communication because people are expected to use language to demonstrate their understanding of social context and their capacity to communicate with others (Wang et al., 2020). For older adults, the deterioration of language function as they age is an important indicator of cognitive decline. In the CGSS questionnaire, cognitive ability is measured by four items: the ability to listen to Mandarin, the ability to speak Mandarin, the ability to listen to English, and the ability to speak English. Each item was measured into the options: ‘do not understand at all ( = 1)’, ‘poorly know ( = 2)’, ‘moderate ( = 3)’, ‘good ( = 4)’, and ‘excellent ( = 5)’. The items were also summed up. The third is social participation, measured by the frequency of seven social activities including going out to the movies, shopping, attending cultural events, gathering with relatives who do not live together, meeting with friends, participating in sports, and watching sports live. The options ranged from ‘never ( = 1)’, ‘several times a year ( = 2)’, ‘several times a month ( = 3)’, ‘several times a week ( = 4)’, to ‘every day ( = 5)’. We performed a Principal Component Analysis to formulate a social participation factor (2013 CGSS: Bartlett test: ${\chi }_{21}^2={1087.008}^{\ast \ast \ast },KMO=0.684$; 2021 CGSS: Bartlett test: ${\chi }_{21}^2={865.253}^{\ast \ast \ast },KMO=0.679$). A higher value shows more engagement in social activities and the associated successful aging.

Pollution awareness

Air pollution awareness (APA) and water pollution awareness (WPA) were measured by whether the respondent was aware of air pollution and water pollution. The categories were constructed as follows: 0 = No, I’m not, 1 = Yes, I’m aware.

Table 1 provides a summary of the measurements of the main variables in the study’s questionnaire.

Table 1. Questionnaire for measuring the main variables at the individual level.

Scale	Subscale	Measure item
Successful aging	Subjective health	On a scale of 1 to 5, what do you think of your current physical condition?
	Cognitive ability	On a scale of 1 to 5, what do you think of your ability to listen to Mandarin, speak Mandarin, listen to English, and speak English?
	Social participation	In the past year, how often do you spend your free time going out to the movies, going shopping, attending cultural events (such as concerts, shows, and exhibitions), gathering with relatives who do not live together, meeting with friends, participating in sports, and watching sports live?
Pollution awareness	Air pollution awareness	Are you aware of air pollution?
	Water pollution awareness	Are you aware of water pollution?

Pollution condition

The macro-level variables of air pollution conditions (APC) and water pollution conditions (WPC) were extracted from China Statistical Yearbook in provinces. APC was measured by the emissions of primary pollutants (sulfur dioxide, nitrogen oxides, and particulate matter) in each region, and the original unit was million tons. We added the emissions of the three pollutants and then calculated the logarithmic values, aiming to balance the bias of the regression estimates due to the regional large values. WPC was also measured by the emissions of major pollutants (chemical oxygen demand, ammonia nitrogen, and total nitrogen) in each region, and the raw scores were transferred using logarithmic values.

Control variables

Social demographic variables selected in the analysis included gender, age, education, residence (coded as rural or urban according to respondents’ households), marriage status (coded as in marriage or not in marriage), number of children, annual family income (log-transformed), social security, and media use. Access to social security was also transcribed into ‘social secured’, as long as the participant was enrolled in one of the social security programmes: basic urban medical insurance, basic rural cooperative medical insurance, public medical insurance, basic urban/rural pension insurance, commercial medical insurance, commercial pension insurance. Media use was measured by one question: ‘In the past year, what was your usage of newspaper, magazine, broadcast, television, Internet, and customised messages for cell phones?’ on a five-point Likert-type scale from ‘never’ ( = 1), to ‘always’ ( = 5) with higher final summed values indicating more frequent media usage.

Analytic strategies

We performed statistical analyses in Stata 17.0. Firstly, we analysed the distribution of critical variables for the older population.

Secondly, we used a seemingly uncorrelated regression model (SUR) for the estimation, which is a joint estimation of the three OLS models. Considering that the dependent variables contained three dimensions and the independent variables were consistent, if three ordinary least squares models (OLS) were estimated separately, factors not incorporated in the models or not observable would exist that affect the three dependent variables at the same time, and three separate models would have inter-correlated residual terms and correlated dependent variables. The use of SUR models can not only improve the efficiency of estimation but also correct the estimation bias caused by the omitted variable problem to some extent (Zellner, 1962). The formula is: $\{\begin{array}{c}Sujective\mathrm{\_}healt{h}_i=\alpha +\sum {\beta }_k{X}_{ik}+{\beta }_k{Z}_{ik}+{\epsilon }_i\\ Cognitive\mathrm{\_}abilit{y}_i=\gamma +\sum {\phi }_k{X}_{ik}+{\phi }_k{Z}_{ik}+{\theta }_i\\ Social\mathrm{\_}participatio{n}_i=\tau +\sum {\rho }_k{X}_{ik}+{\rho }_k{Z}_{ik}+{\mu }_i\end{array}$where $X_{ik}$ denotes the k-variable of individual $i$ and $Z_{ik}$ denotes the control variables. $\alpha$, $\gamma$, and $\tau$ represent the intercept of the three models. ${\epsilon }_i$, ${\theta }_i$, and ${\mu }_i$ are random error terms.

Thirdly, we used multilevel models to examine the differential effects of regional macro variables on individuals, and random-coefficient regressions were set up as follows: $Y_{ik}={\delta }_{0k}+{\delta }_{1k}{X}_{ik}+{\delta }_z{Z}_{ik}+{\omega }_{ik}$ ${\delta }_{0k}={\sigma }_{00}+{\sigma }_{01}pollution\mathrm{\_}condition{s}_k+{\vartheta }_{0k}$ ${\delta }_{1k}={\sigma }_{10}+{\sigma }_{11}pollution\mathrm{\_}condition{s}_k+{\vartheta }_{1k}$

where ${\omega }_{ik}$ represents the individual error term.

Finally, we employed propensity score matching analysis (PSM) to respond to the possible deviations in estimated coefficients due to selection bias, Nearest-neighbor matching within calliper (Smith, 1997) and Kernel matching (Heckman et al., 1997), were applied.

Results

Table 2 shows the mean values and standard deviations of the key analytic variables. Figure 2 illustrates the spatial distribution of pollution. Compared with the samples in 2013, samples from 2021 had a slightly lower degree of subjective health and cognitive ability, but they were more active in social participation. More than 80 percent of samples were aware of air pollution and water pollution in 2013, in comparison to 65 percent found in the 2021 survey. The pollution conditions were improved in 2021 and the total amount of pollutants was decreasing.

Figure 2. Air and water pollution conditions in 2013 and 2021.

Notes. 1. The maps are based on China’s provincial districts. 2. Air and water pollutants are summed up by three major pollutants separately. The darker the colour of the area, the more serious the pollution. Areas without pollution data are shown in white.

Table 2. Descriptive statistics of studied variables.

Variables	Mean (SD)
	2013 (N = 5404)	2021 (N = 1580)
Dependent variables
Subjective health (1–5)	3.32 (1.10)	3.16 (1.11)
Cognitive ability (4–20)	8.54 (2.33)	8.39 (2.37)
Social participation	0 (1.00)	0.03 (0.99)
Independent variables
Air pollution awareness (0,1)	0.86 (0.35)	0.66 (0.47)
Water pollution awareness (0,1)	0.87 (0.34)	0.65 (0.48)
Macro-level variables
Air pollution condition	5.09 (0.60)	3.91 (0.67)
Water pollution condition	4.56 (0.70)	4.46 (0.92)
Control variables
Gender (Female = 0)	0.52 (0.50)	0.47 (0.50)
Age	63.04 (9.29)	64.64 (9.70)
Education (0–19)	6.91 (4.62)	7.43 (4.34)
Residence (Rural residence = 0)	0.46 (0.50)	0.39 (0.49)
Marriage (Not in marriage = 0)	0.80 (0.40)	0.77 (0.42)
Number of children	2.34 (1.39)	2.10 (1.25)
Social security (Not secured = 0)	0.95 (0.22)	0.97 (0.18)
Media usage (6–30)	12.34 (3.88)	12.97 (4.05)
Annual family income	44,256.25 (48,501.22)	92,184.12 (49,4085.9)

Focusing on the extent to which older adults were aging successfully, we found that only 15.16 percent of older adults in 2013 and 11.58 percent in 2021 had higher values of subjective health, cognitive ability, and social participation at the same time than the mean value, indicating that most older adults were not in good physical and social functioning conditions. The percentage was similar to other academic estimations that about 35 percent of older people in developed countries are thriving and 14 percent in developing countries, encouraging the adoption of optimistic social policy (Meng & D’Arcy, 2014). The group with advantaged social status had higher values of subjective health, cognitive ability, and social participation, which means male older adults with urban residence, in marriage, and social security had better conditions to age successfully than female rural residents, not in marriage, and not covered by social security (see Table 3).

Table 3. The mean values of successful aging in social-demographic groups.

SA	2013	2021
Subjective health	Female = 3.232, Male = 3.403	Female = 3.071, Male = 3.257
	Rural = 3.218, Urban = 3.439	Rural = 3.085, Urban = 3.274
	Not married = 3.098, In marriage = 3.377	Not married = 3.043, In marriage = 3.194
	Not secured = 3.167, Secured = 3.329	Not secured = 3.000, Secured = 3.164
Cognitive ability	Female = 8.360, Male = 8.700	Female = 8.239, Male = 8.554
	Rural = 7.625, Urban = 9.594	Rural = 7.635, Urban = 9.572
	Not married = 7.942, In marriage = 8.690	Not married = 8.022, In marriage = 8.500
	Not secured = 8.407, Secured = 8.542	Not secured = 7.944, Secured = 8.403
Social participation	Female = −0.014, Male = 0.013	Female = −0.028, Male = 0.088
	Rural = −0.335, Urban = 0.390	Rural = −0.188, Urban = 0.364
	Not married = −0.103, In marriage = 0.024	Not married = −0.101, In marriage = 0.066
	Not secured = −0.225, Secured = 0.012	Not secured = −0.245, Secured = 0.036

We performed SUR models to estimate the regression results (see Table 4). From the Breusch–Pagan Test of Independence in the pretest for the SUR models, the results rejected the hypothesis of independent perturbation terms in three equations at the 0.1 percent level $({\chi }_{2013}^2=120.87,p=0.000{\chi }_{2021}^2=49.54,p=0.000)$, which means the three dimensions of SA were internally correlated and the usage of SUR models could capture the inner trait. In the 2013 models, compared with people with no awareness of air pollution, people knowing about air pollution had better subjective health and cognitive ability $({\beta }_{subjective\mathrm{\_}health}=0.145,p<0.05{\beta }_{cognitive\mathrm{\_}ability}=0.332,p<0.01)$. Compared with people without awareness of water pollution, people aware of it had a significantly higher level of subjective health and social participation $({\beta }_{subjective\mathrm{\_}health}=0.138,p<0.05{\beta }_{social\mathrm{\_}participation}=0.131,p<0.05)$. In the 2021 models, the correlation between APA, WPA, and subjective health was significant and positive. Comparing two years, the effects of personal awareness of air pollution on SA surpassed the effects of water pollution for the significant coefficients were modestly higher. These findings partially supported Hypotheses 1 and 2. Cumulatively, this model explained about 37 percent of the differences across older persons’ cognitive ability, 20–28 percent of social participation, and 6–9 percent of subjective health.

Table 4. SUR models: the relationship between environmental awareness and SA.

	2013 (N = 5377)			2021 (N = 1572)
	Model 1-1	Model 1-2	Model 1-3	Model 2-1	Model 2-2	Model 2-3
Individual level	Subjective health	Cognitive ability	Social participation	Subjective health	Cognitive ability	Social participation
APA	0.145* (0.066)	0.332** (0.116)	−0.024 (0.053)	0.256** (0.079)	0.030 (0.139)	0.034 (0.065)
WPA	0.138* (0.067)	0.128 (0.118)	0.131* (0.054)	0.174* (0.078)	0.367 (0.136)	0.048 (0.063)
Male	0.079** (0.030)	−0.212*** (0.053)	−0.129*** (0.024)	0.123* (0.057)	−0.229* (0.010)	−0.013 (0.047)
Age	−0.018*** (0.002)	−0.022*** (0.003)	−0.006*** (0.002)	−0.012*** (0.003)	−0.020** (0.006)	−0.003 (0.003)
Education	0.008* (0.004)	0.141**** (0.008)	0.009** (0.003)	0.009 (0.008)	0.178*** (0.015)	0.014* (0.006)
Urban residence	−0.003 (0.035)	0.798*** (0.063)	0.086 (0.067)	0.572** (0.175)	0.802*** (0.117)	0.228*** (0.055)
In marriage	0.030 (0.038)	0.090 (0.067)	−0.035 (0.031)	0.004 (0.068)	−0.094 (0.120)	0.027 (0.056)
No. of children	−0.021 (0.013)	−0.062** (0.007)	−0.015 (0.010)	−0.072** (0.024)	−0.120** (0.043)	−0.025 (0.020)
Annual family income	0.018*** (0.004)	0.019** (0.007)	−0.010** (0.003)	0.007 (0.006)	0.020^+ (0.011)	0.004 (0.005)
Social secured	0.089 (0.065)	−0.072 (0.115)	0.152** (0.052)	0.018 (0.151)	−0.430 (0.266)	0.086 (0.124)
Media use	0.031*** (0.005)	0.125*** (0.118)	0.104*** (0.004)	0.019* (0.008)	0.093*** (0.014)	0.080*** (0.007)
Intercept	3.492*** (0.151)	6.742*** (0.266)	−1.156*** (0.122)	3.654*** (0.286)	7.249*** (0.502)	−1.142*** (0.234)
R-squared	0.09	0.37	0.28	0.06	0.37	0.20
${\chi }^2$	539.68***	3210.72***	2142.67***	98.26***	916.47***	402.69***

Note: Significance levels: ⁺p < 0.1, *p < 0.05, **p < 0.01, ***p < 0.001 (two-tailed test). All the results in the paper are interpreted with the same standard.

As for control variables, men were much less cognitively capable than women, despite having higher subjective health. People would age less well as they get older since all three indicators were declining. Higher levels of cognitive ability and social participation were found among older people who had longer years of schooling. Urban residents had significantly greater levels of SA than rural residents, which highlights the contrasting consequences of residential status in 2021. Media use was significantly correlated with subjective health, cognitive function, and social engagement, suggesting that greater media use was associated with higher levels of SA.

Table 5 demonstrates the MLM results in 2013 to observe the moderating effects of regional pollution status between individual pollution perception and SA. We performed the Intraclass Correlation Coefficient (ICC) in the null model first: $IC{C}_{subjective\mathrm{\_}health}=0.080$, $IC{C}_{cognitive\mathrm{\_}ability}=0.266$, $IC{C}_{social\mathrm{\_}participation}=0.183$，which meant that provincial-level variables accounted for 8.0 percent to 26.6 percent of SA differences, necessitating the employment of MLM to explain the influence of factors at various levels.

Table 5. MLM: environmental awareness and SA, under the influence of regional pollution conditions (2013, N = 5377, Group = 28).

	Model 3-1	Model 3-2	Model 3-3	Model 3-4	Model 3-5	Model 3-6
	Subjective health	Cognitive ability	Social participation	Subjective health	Cognitive ability	Social participation
Individual level
APA	0.267 (0.593)	2.563* (1.062)	1.196** (0.433)	0.088 (0.065)	0.247* (0.107)	0.009 (0.052)
WPA	0.135* (0.066)	0.207^+ (0.108)	0.123* (0.053)	0.287 (0.408)	1.823* (0.731)	0.560^+ (0.318)
Control variables	Controlled	Controlled	Controlled	Controlled	Controlled	Controlled
Intercept	2.877*** (0.666)	6.513*** (1.474)	−1.988*** (0.157)	2.903*** (0.557)	7.274*** (0.336)	−1.317** (0.491)
Provincial level
APC	0.110 (0.162)	−0.356 (0.405)	0.102 (0.140)	0.085 (0.127)	−0.712^+ (0.368)	−0.099 (0.120)
WPC	0.030 (0.110)	0.475 (0.306)	0.060 (0.098)	0.052 (0.124)	0.702* (0.322)	0.138 (0.113)
Interaction
$\mathrm{APA}\times \mathrm{APC}$	−0.033 (0.113)	−0.446* (0.202)	−0.228** (0.082)
$\mathrm{WPA}\times \mathrm{WPC}$				−0.031 (0.086)	−0.350* (0.155)	−0.093 (0.067)
Log-likelihood	−7758.71	−10434.72	−6592.44	−7758.91	−10433.27	−6594.66

As for the individual level, older people aware of air pollution had higher levels of cognition and social participation than people without awareness of air pollution $({\beta }_{cognitive\mathrm{\_}ability}=2.563,p<0.05{\beta }_{social\mathrm{\_}participation}=1.196,p<0.01)$. Water pollution awareness yielded similar effects $({\beta }_{cognitive\mathrm{\_}ability}=1.823,p<0.05{\beta }_{social\mathrm{\_}participation}=0.560,p<0.1)$. Moreover, the interaction effects of APA and APC were statistically significant at the dimensions of cognitive ability and social participation $({\beta }_{cognitive\mathrm{\_}ability}=-0.446,p<0.05{\beta }_{social\mathrm{\_}participation}=-0.228,p<0.01)$, indicating that the impact of regional air pollution on SA varied depending on whether or not a person was aware of pollution awareness. In areas with high regional pollution compared to areas with low pollution, the connection between pollution awareness and SA was less. The protective role of environmental awareness was also illustrated here, as people were likely to take actions to protect the environment and maintain health as long as they were environmentally aware, despite the deterioration of the actual environment. As for the interaction of WPA and WPC, the effect was significant for impaired cognition $({\beta }_{cognitive\mathrm{\_}ability}=-0.350,p<0.05)$ but not for social engagement. We deduced that air pollution was easier to perceive and more likely to affect the outdoor activities of older adults than water pollution, thus showing significant effects. The effects on subjective health were parallel yet failed the significance test.

To understand the patterns of regional effects on individuals, Figure 3 represents predicted SA extents by regional pollution conditions. The predictions were based on the significant results in Table 5, Model 3-2, 3-3, and 3-5. Figure 3-1 shows that older people’s cognitive function declined with declining air quality, and those who perceived air pollution suffered from significantly greater cognitive deterioration. Figure 3-2 shows the heterogeneity of social participation of older age groups as influenced by air pollution status and perception. Air pollution had a little social impact on older persons who did not experience it, but it significantly reduced social participation in those who did. Figure 3-3 presents the intriguing finding that in places with high levels of water pollution, older persons who perceived pollution had lower levels of cognition than older adults who did not experience pollution. The marginal effects also explained why macro variables were sometimes insignificant as there is a possibility that the positive and negative effects generated by environmental awareness could be offset.

Figure 3. Predicted successful aging by personal environmental awareness and provincial environmental conditions with 95% Confidence Intervals (the figures are listed as Figure 3-1, 3-2, and 3-3 from up to down).

Table 6 displays the MLM results in 2021. Given the limitations of the sample size and grouping in 2021, we used multilevel models for cognitive ability and provincial fixed-effects models for health perception and social participation to meet the prerequisites for model estimation and to achieve a better fit. The interaction terms revealed that the interaction effects between individual perceptions of air pollution and regional air pollution levels were significantly negative in the SA dimensions of subjective health and social participation in the 2021 sample $({\beta }_{subjective\mathrm{\_}health}=-0.295,p<0.01{\beta }_{social\mathrm{\_}participation}=-0.161,p<0.1)$. This suggested that the influential degree of individual air pollution awareness for SA eventually declined as regional air pollution increased. The decreasing effect of air pollution perception on SA was less pronounced than the interaction effect in 2013. The failure of significance tests in the interaction effects between personal perceptions of water pollution and pollutants’ amount suggested that, in contrast to air pollution, older people’s awareness of water pollution may not be sensitive enough to influence the SA process.

Table 6. MLM: environmental awareness and SA, under the influence of regional pollution conditions (2021, N = 1572, Group = 19).

	Model 4-1	Model 4-2	Model 4-3	Model 4-4	Model 4-5	Model 4-6
	Subjective health	Cognitive ability	Social participation	Subjective health	Cognitive ability	Social participation
Individual level
APA	0.891^+ (0.475)	−0.886 (0.917)	0.669^+ (0.390)	−0.294*** (0.378)	−0.039 (0.134)	0.015 (0.066)
WPA	0.181* (0.077)	0.166 (0.129)	0.063 (0.063)	0.134 (0.336)	−0.148 (0.656)	0.111 (0.276)
Control variables	Controlled	Controlled	Controlled	Controlled	Controlled	Controlled
Intercept	6.364 (4.991)	9.251*** (1.287)	−3.172 (4.104)	6.955 (5.010)	8.862*** (1.171)	−2.930 (4.115)
Provincial level
APC	1.151 (1.158)	−0.087 (0.431)	−0.648 (0.952)	−0.294*** (0.127)	0.024 (0.392)	−0.766 (0.951)
WPC	−0.078 (0.098)	−0.222 (0.308)	1.151 (1.983)	0.953 (1.158)	0.236 (0.307)	1.209 (1.987)
Interaction
$\mathrm{APA}\times \mathrm{APC}$	−0.295** (0.115)	0.206 (0.222)	−0.161^+ (0.095)
$\mathrm{WPA}\times \mathrm{WPC}$				0.009 (0.073)	0.071 (0.143)	−0.012 (0.060)
Log-likelihood	\	−3447.29	\	\	−3169.96	\
F-value	3.56***	\	14.92***	5.15***	\	14.80***

Overall, it appeared that the MLM analysis findings followed similar patterns between 2021 and 2013, that is, pollution awareness moderated the effect of pollution conditions on SA and air pollution awareness had a notable effect.

Since older adults who perceived environmental contamination may be highly selective, and this selection bias may induce distortion in the estimated coefficients, PSM analysis was used to mitigate this problem to some extent. Using the two methods of Nearest-neighbor matching within calliper (with 0.01 standard deviation) and Kernel matching, we showed the Average Treatment Effect of the Treated (ATT) for the intervention group (see Table 7). The PSM results were generally robust considering the previous SUR results. Compared with older people with no APA and WPA, people aware of pollution had constantly higher values of subjective health, cognitive ability, and social participation. The statistical significance of the difference between the two groups in 2021 was reduced compared to the difference in 2013 between the perceived and unperceived groups for environmental pollution.

Table 7. PSM analysis.

Variables	APA: yes	APA: no	SD (of Difference)	T-value	WPA: yes	WPA: no	SD (of Difference)	T-value
2013
Nearest-neighbor matching within calliper-0.01SD
Subjective health	3.163	2.916	0.059	4.21	3.236	2.915	0.060	5.33
Cognitive ability	7.797	7.182	0.108	5.70	7.853	7.276	0.113	5.13
Social participation	−0.252	−0.383	0.046	2.85	−0.259	−0.410	0.046	3.32
Kernel matching
Subjective health	3.089	2.854	0.074	3.18	3.381	3.151	0.059	3.93
Cognitive ability	8.674	8.006	0.110	6.07	8.709	8.200	0.113	4.53
Social participation	0.028	−0.097	0.050	2.52	0.052	−0.129	0.047	3.88
2021
Nearest-neighbor matching within calliper-0.01SD
Subjective health	3.114	3.223	0.059	−1.56	3.176	3.138	0.069	0.55
Cognitive ability	8.203	7.984	0.137	1.60	8.319	8.103	0.143	1.50
Social participation	0.018	−0.110	0.059	2.17	0.037	−0.065	0.060	1.71
Kernel matching
Subjective health	3.131	3.266	0.064	−2.11	3.174	3.151	0.061	0.38
Cognitive ability	8.606	8.365	0.129	1.87	8.569	8.277	0.124	2.35
Social participation	0.094	0.046	0.054	0.90	0.083	0.002	0.052	1.55

Discussion

Results and limitations

We summarised the empirical findings and arrived at the following key conclusions: firstly, the degree of subjective health and cognitive capacity was strongly impacted by air pollution awareness, whereas subjective health and social involvement were significantly impacted by water pollution awareness. Environmental awareness might prompt actions to acquire healthy knowledge and improve health conditions. Being aware of pollution is considered to be a key indicator of one’s understanding of and support for resolving ecological issues, as well as one’s willingness to contribute to those efforts and lessen their heinous impacts (Dienes, 2015). Compared to earlier studies that concentrated on persons of working age (Ramírez et al., 2019), the protective effect of environmental awareness on personal life was also important for older adults.

Secondly, the combined effects of individual and regional factors indicated cumulatively detrimental influences of environmental consciousness and circumstances on SA. Environmental determinants were conditionally lessening their effects for older adults with pollution awareness. The analysis of environmental influences on successful aging was extended by this finding, which combined objective conditions with subjective perceptions. It showed that the adverse effects of environmental pollution were not always severe and that the protective effects of environmental awareness could sometimes moderate the impact of pollution on older adults’ cognitive and social participation.

Thirdly, the impacts of air pollution awareness outweighed those of water pollution awareness in two survey times. The lowering impacts of water pollution circumstances were not apparent due to the distinct influence of air quality and the subtle influence of water quality, but this did not indicate that water pollution influences have greater prospects than air pollution. On the contrary, water pollution is a growing challenge, and calls for attention to integrated awareness, knowledge, and understanding (Awoke et al., 2016). A difficult first step in reducing the prevalence of frailty in older people’s social and physical functions is acknowledging the severity of various contaminants.

The study has several limitations. The first is that we did not use mediation analysis to identify the underlying mechanisms between environmental factors and SA. Second, although the original survey data were focused on assessing linguistic skill as a cognitive capacity, it might overlook other elements of abstraction, orienting, reasoning, and calculating to demonstrate a thorough examination of cognition. The insufficient comparison of time sequence is the third. The final one is that we only established a correlation between environment and aging, not a causal relationship, which may be enhanced by applying sophisticated statistical models and accounting for potential endogenous issues.

Research and policy implications

This study has advanced the knowledge of the relationship between environmental factors and SA among Chinese older adults. Specifically, we examined the joint effects of environmental pollution on the three facets of aging, which were subjective health, cognitive ability, and social involvement. Second, we separated the various effects of air and water pollution, illuminating the diversity of various forms of environmental pollution that had an impact on aging. Third, we examined the aging of China at two-time points and contrasted them to contribute to the understanding of Chinese SA policy advancement.

A significant concern for China’s old-age, health, and environmental protection policies is the conundrum of the aging dilemma and environmental issues (Cohen & Gerber, 2017). Understanding how environmental pollution interferes with successful aging can help provide policy suggestions for improving environmental qualities. In particular, at a time when economic growth continues to be a top policy objective, finding the optimal balance between environmental pollution and economic development to enhance the quality of the elderly population’s outdoor living environment has more immediate and long-term practical significance (Chen et al., 2013). Indeed, the success of environmental protection during the 2008 Summer Olympics and 2022 Winter Olympics in Beijing shows the potential for local government pollution control, which indicates that the rising importance of pollution prevention in the governmental evaluation system will bring giant implementation impacts (Zhang et al., 2010). Adhering to the existing policies, new environmental policies should be formulated to target specific and complex air and water pollution problems, such as the spatial spillover effects caused by the regional transfer of polluting industries (Zhang et al., 2022). Also, environmental laws and regulations should be integrated, monitored, and supported by incentives like tax breaks to encourage industrial companies to employ clean energy and modernise their technology. To solve environmental inequality, there is a call for developing external environmental risk management organisations, strengthening their accountabilities, and promoting cooperation and collaborations (Zhang et al., 2023). At the same time, a national platform for pollution information should be established to share monitoring data, and adequately enhance environmental pollution control standards in areas with a high elderly population.

Furthermore, it also suggests guidelines for empowering the older population to lead healthy lives, enhance their quality of life, and engage in social activities. The usage of cell phones, television, the Internet, and other media can be designed to educate locals, particularly the elderly, about the environment so that they can fully understand and grasp environmental knowledge and know they have a responsibility to protect the environment. This will enable them to consciously participate in environmental protection actions and increase social participation in the process, keeping their bodies functioning healthily. Besides, the removal of institutional barriers and the ongoing reform of medical systems would help the underprivileged population in obtaining care resources and social support (Feng et al., 2015). The provision of a favourable climate for constructing an age-friendly society is regarded as a catalyst for enhancing the well-being and engagement of the aging population.

Acknowledgments

The authors would like to thank Xiaoguang Li, Qian Zhang, and Yixue Zhang, the researchers at the Institute for Empirical Social Science Research at Xi’an Jiaotong University, for their valuable discussions and comments on the manuscript. We are also grateful to the Editors and anonymous reviewers for their revision suggestions.

Disclosure statement

No potential conflict of interest was reported by the author(s).