The Influence Mechanism of the Community Subjectively Built Environment on the Physical and Mental Health of Older Adults

Abstract

In order to clarify the mechanism by which subjectively built environments impact the physical and mental health of older adults and promote the construction of “healthy aging” and “healthy cities,” this study develops a structural equation model based on questionnaire data from older adults in Hefei and constructs a mechanism of a “community subjectively built environment—physical and mental health” with leisure physical activities and social interaction activities as mediators. The results indicate that the specific combination of subjectively built environmental factors such as community safety and security, internal supportive living facilities, a green environment, a walking environment, and a degree of beautification significantly impacts the physical and mental health of older adults. Leisure physical activity and social interaction activities play different roles in mediation, forming two sets of action mechanisms: “community-built environment—leisure physical activity—physical health” and “community-built environment—social interaction activity—physical and mental health.”

1. Introduction

“Healthy aging” is one of the least costly and most effective ways of dealing with population aging [1]. However, the rapid urbanization and motorization process poses a serious threat to “healthy aging” by changing the living environment and lifestyle of older adults. Studies based on the objectively built environment have found that the urban built environment significantly affects residents’ physical and mental health and weight [2,3,4,5]. However, few studies have explored how the subjectively built environment is associated with residents’ physical and mental health [6,7]. The health problems of older adults are becoming increasingly prominent as the population ages. This process will seriously affect the quality of life of older adults and increase the economic burden on the state, society, and families. It will also significantly reduce the ability of older adults to respond to public health emergencies [8]. Urban planners are eager to gain a deeper and more thorough understanding of the relationship between the built environment and the health of older adults to promote healthy aging.

The built environment usually affects the health of residents not directly but indirectly through health-related behaviors [9], such as physical activity, dietary behavior, and social interaction [10,11,12,13]. However, the pathways of action for the built environment’s effects on health are far from reaching a consensus, and more evidence is needed. In particular, most previous empirical studies have chosen one mediator for a single mediating effect analysis. However, the same built environment element may affect an individual’s health differently due to different mediators [14]. Overall, planners have little knowledge of how the built environment affects the health of older adults and what exactly the mechanisms are.

The self-selection effects of residents mean that individuals choose to live in communities with different built environmental characteristics based on their attitudinal preferences and behavioral habits. Hence, their physical and mental health may be caused by the built environment they live in or may be related to self-selection [15]. The self-selection effects of residents have been confirmed in Chinese and foreign urban studies, but many studies still do not consider them [16]. Ignoring the effects of self-selection is likely to misestimate the built environment’s impact on individual health and thus mislead the development of healthy urban planning policies [17].

Due to these research shortcomings, this study aims to answer two research questions. First, what are the specific pathways by which the subjectively built environment of the community affects the physical and mental health of older people? Second, after considering the self-selection effect of residence, will the path by which the subjectively built environment of the community affects the health of older adults change? To answer these questions, we conducted a sample survey of older adults (residents aged 55 and above) in Hefei, Anhui Province, a city with an increasingly aging population. Based on the questionnaire data, we built a structural equation model (SEM) with physical activities and social interaction activities as mediators and personal attributes as moderating variables to clarify the mechanism of the influence of the subjective community-built environment on the self-rated physical and mental health of older adults. Then, the robustness of the research results was tested using China’s housing system to control the self-choice behavior of older adults.

This study makes a few contributions to the literature. First, it explores the impact of multiple subjectively built environmental elements on the health of older people through combinations. This approach is an essential addition to previous studies that have only explored the independent effects of individually built environmental elements on health, and it helps to improve the effectiveness of healthy city planning implementation [18]. Second, circumventing the influence of residential self-selection through China’s unique urban housing system (the welfare housing distribution system before 1998 and the social security housing system after the market reform have left residents in a passive distribution position, with almost no self-selection effect in terms of residence) helps to clarify whether health outcomes for older people are a function of the built environment or resident preferences, providing a new perspective for precise interventions in healthy urban planning. Finally, by observing the mediating variables of leisure physical activity and social interaction activity, the built environment’s impact on older people’s health can be explained in greater depth than traditional attribution analysis. This helps to reveal the mechanisms of the built environment’s impact on older people’s health, identifying directions for planning interventions and providing an important and innovative way to integrate health concepts into planning and design.

The remainder of this paper is organized as follows. Section 2 provides a literature review and proposes the research hypotheses. Section 3 introduces materials and methodology, including the study area, conceptual framework, data, variables, and methods. Section 4 presents the results, and we discuss the research findings and analyze the theoretical contributions in Section 5. Section 6 concludes with the main findings and shortcomings of our research and the direction for future studies.

2. Literature Review and Research Hypotheses

Optimizing the built environment is one of the most feasible ways to improve residents’ physical and mental health [14]. The health effect of the community-built environment is more significant for older people who spend long periods in residential conditions, where a well-built environment can play a positive role in the health of older people by promoting individual physical activities and social interaction activities [8]. Compared with passive medical technology, the effect of initiative optimization of the built environment to create a living environment conducive to physical and social interaction activities has many advantages, such as durability, universality, and economy [19,20]. The existing empirical research shows that a compactly built environment, that is, a built environment with high density, a high function mix, and high road connectivity, can effectively control weight, reduce the risk of obesity, and improve the physical and mental health of residents [21,22]. This is because a compactly built environment shortens travel distances and promotes the adoption of active modes of travel such as walking and cycling, which in turn increases physical activity and creates more opportunities for people to meet informally [21,22,23,24,25]. As Jane Jacobs notes in The Death and Life of Great American Cities, density and mixing make cities safer and more pleasing to the eye [26]. However, there are also some studies showing that a compactly built environment reduces living privacy, increases the flow of people and vehicles [27], and produces air and noise pollution [28], which negatively impacts the physical and mental health of residents [15,28,29].

However, most studies about the association of the built environment with health are based on the objectively built environment, which may not reflect the subjective environment. While objectively built environment indicators have the advantage of being easy to quantify, standardize, and translate research findings into policy, they also have two disadvantages [6,30]. On the one hand, it is difficult to fully capture the health effects of the objectively built environment because its indicators may not reflect the authentic environmental exposure of the older adult population due to geographical and contextual uncertainty [31]. On the other hand, the same objectively built environment may have different health effects on different people because different people may have different perceived outcomes for the same facilities and environment [32]. However, due to the impact of individual attributes, actual use, and other factors, the subjectively built environment has the advantage of being closer to the true psychology and behavior of older adults, so it has unique health effects that can reflect the health effects that cannot be captured by the objectively built environment [31,33,34].

Considering that the built environment often does not directly affect individual health, many studies have used different health-related behaviors as mediators to prove the pathway of the impact of the built environment on individual health [35]. Most studies have regarded physical activity, social interaction activity, and eating behavior as mediators when discussing the mechanism of the built environment’s influence on individual health [36]. In Chinese cities, Gao et al. [37] and Wu et al. [38] observed that physical activity significantly mediated the impact of the built environment on the health of older adults. By contrast, Yu et al. [18] found that compared with physical activity, social interaction activities played a more significant mediating role in revealing the effect of built environment dimensions such as density, public facilities, medical accessibility, and location on health. Zhang et al. [14] used social interaction activity, physical activity, and eating behavior as mediators to examine the complete pathway of the effect of the built environment on individual health. Although many studies have empirically examined mechanisms of built environment effects on health, these results are mixed and pay little attention to the subjectively built environment.

Urban studies at home and abroad have demonstrated that the built environment’s impact on individual health may be biased by residential self-selection [39]. To address the issue of residential self-selection and to explore the net effect of the built environment on health, Cao et al. [40] point out that relocation housing provides a good natural experiment for exploring this relationship, an approach also adopted by both Sun et al. [41] and Zhang et al. [14] Alternatively, in the absence of self-selection data, propensity score models can address confounding due to residential self-selection, as was done in some previous studies [15,24,42].

In this study, we aim to explore the mechanisms of the community’s subjectively built environment on the physical and mental health of older adults and test the robustness of the empirical results by controlling self-selection. We propose two hypotheses.

H1. 

The subjectively built environment of the community has an indirect effect on the physical and mental health of older people through leisurely physical and social interaction activities.

H2. 

The results of the empirical tests are broadly robust after controlling for the residential self-selection of older people.

3. Materials and Methodology

3.1. Study Area

As a rising provincial capital in central China, Hefei, Anhui Province, is comprehensively enhancing its function as a world-class sub-center city in the Yangtze River Delta and accelerating its development into an emerging international megacity (Figure 1). In 2021, the city’s gross domestic product (GDP) reached CNY 114.28 billion, ranking 19th among large and medium-sized cities in China. However, Hefei is the first city in Anhui Province to enter an aging society. By the end of 2021, the city’s population aged 60 years and above was 1.427 million, accounting for 15.08 percent of the total population, and the population aged 65 years and above was 1.177 million, accounting for 12.44 percent of the total population. Compared with China’s sixth national census in 2010, the number of people aged 65 and above in Hefei increased by 744,000, while the proportion of people aged 65 and above increased by 4.27%. With the continuous aging of the population, the proportion of older adults in Hefei will continue to gradually increase. The growing number of older adults means that Hefei is facing massive social and economic expenditures related to the health of older adults. Aging has become a vital public health and social problem in Hefei at present and in the future for a considerable period of time. Hefei is currently undergoing rapid urbanization, with the urban population accounting for 84.4% of the total population in 2021, an increase of 21.2% compared to 2010, and the urban built-up area reaching 502.5 square kilometers in 2020, an increase of 44.3% compared to 2010. The urban space development of Hefei city is at a critical stage of functional transformation and reconstruction, which will indirectly impact the health of older adults. Therefore, studying the relationship between the built environment and the health of older people in Hefei is typical and representative.

3.2. Data

The data for this study are mainly from questionnaires and interviews. The social survey was conducted in July 2021, covering subjectively built environments, social interaction activities, physical activities, physical health, mental health, and individual characteristics. The study adopted a stratified sampling method to select 11 representative neighborhoods in Yaohai District, Shushan District, Baohe District, and Luyang District of Hefei for questionnaire surveys and interviews based on characteristics such as spatial location, community type, year of construction, and surrounding supportive facilities (Figure 1). A total of 535 questionnaires were returned, and 505 valid samples were obtained after eliminating respondents with missing variables by tabulation deletion (a method of deleting the sample whenever there is a missing value).

Table 1. Descriptive statistics of individual characteristics.

Manifest Variable	Description	Percentage
Length of residence	Less than 3 years = 1	10.6%
	3–5 years = 2	9.8%
	5–10 years = 3	16.8%
	More than 10 years = 4	62.8%
Household registration	Urban household registration = 0	77.6%
	Rural household registration = 1	22.4%
Local	Yes = 1	77.4%
	No = 2	22.6%
Age	55–64 = 1	28.7%
	65–74 = 2	46.7%
	Over 75 = 3	24.6%
Gender	Male = 0	49.8%
	Female = 1	50.2%
Housing property right	Organization-owned housing, Indemnified apartment = 1	20.4%
	Commercial housing, Rental house, Self-built house = 2	79.6%

shows the individual characteristics of the valid samples. The survey samples are all older adults aged 55 and above, with slightly more males than females. The vast majority are urban residents and local residents of Hefei, and over 60% of the older adults have lived in the survey community for more than 10 years.

3.3. Conceptual Framework

The social–ecological model, the basis of healthy city theory, holds that individual factors, social networks, extensive background conditions, and the environment affect people’s health status. The environmental factors include the built environment, the social environment, and the policy environment [43,44]. When discussing the mechanisms of the built environment on health status, most studies have conducted mediating effects analyses with a single mediator (e.g., physical activities, social interaction activities, dietary behavior, criminal activity) [9]. However, some studies have combined existing fragmentary analyses to conduct multiple mediating effects analyses. For example, Zhang Yanji and co-authors quantified social capital, residential security, physical activities, and dietary behavior as mediators under one framework [14]. Most existing studies control individual characteristics as covariates [45], but the moderating role of individual characteristics has been observed in studies on obesity [46]. Now there is a need to understand how the interrelationship between the subjectively built environment and health may be moderated by individual characteristics [47]. Therefore, this study uses the community-built environment as the independent variable, older people’s health as the dependent variable, leisure physical activities and social interaction activities as the mediators, and individual characteristics of older people as the moderators. These parameters are used to construct a complete SEM of the community-built environmental impact on physical and mental health. The study systematically investigates the mechanism of the community-built environmental impact perceived by older people on their physical and mental health. This approach allows us to reveal the mediating role of leisure physical and social interaction activities and to detect the moderating role of individual characteristics of older people (Figure 2).

3.4. Variables

At the beginning of model construction, safety and security, internal supportive living facilities, a green environment, a walking environment, transportation convenience, a beautification degree, sports and exercise facilities, shopping places, and leisure and entertainment places were selected to represent the subjectively built environment. Meanwhile, neighbor interaction degree, the number of familiar neighbors, neighbor interaction satisfaction, and level of social activity participation were selected to represent social interaction activities. After screening, the variables shown in

Table 2. Variable measurements and descriptive statistics (independent variables, dependent variables, and mediators).

Latent Variables	Manifest Variable	Description	Mean	SD
Subjectively built environment	Safety and security	Very dissatisfied = 1~Very satisfied = 5	3.49	1.00
	Internal supportive living facilities	Very dissatisfied = 1~Very satisfied = 5	3.36	0.97
	Green environment	Very dissatisfied = 1~Very satisfied = 5	3.49	1.02
	Walking environment	Very dissatisfied = 1~Very satisfied = 5	3.64	0.89
	Beautification degree	Very dissatisfied = 1~Very satisfied = 5	3.48	1.02
Social interaction activities	Neighbor interaction degree	Never = 1 ~Frequent = 5	3.30	1.11
	Number of familiar neighbors	0 people = 1; 1–3 people = 2; 4–10 people = 3; 10–20 people = 4; More than 20 people = 5	3.22	1.24
	Neighbor interaction satisfaction	Very dissatisfied = 1~Very satisfied = 5	3.47	1.03
Manifest variable		Description	Mean	SD
Leisure physical activities		Duration of single leisure activity for older adults, Unit: min	96.59	53.06
Physical health		Very dissatisfied = 1~Very satisfied = 5	3.82	1.05
Mental health		Very dissatisfied = 1~Very satisfied = 5	4.02	1.00

were ultimately obtained. The variable filtering process consisted of two steps: the first step was to conduct factor analysis and reliability analysis to remove the variables affecting validity and reliability, and the second step was to remove the variables affecting the fitness of the SEM in the process of building the SEM revision.

3.4.1. Subjectively Built Environment as an Independent Variable

The study selected the subjectively built environments of the community as the independent variables. The objectively built environment indicates the external availability of the built environment but not how much of the built environment is used subjectively by residents. One reason is that people living in a built environment that is highly supportive of activity may have a lower propensity to use it [32,48]. In addition, as older people age, their physical and cognitive functioning declines, and their connections with members of their social networks diminish, leading to a reduction in the scope of travel for older people and the community becoming the main venue for daily activities [8,49]. The study, therefore, selected five observational variables to reflect the subjectively built environment of the community: safety and security, internal supportive living facilities, green environment, walking environment, and beautification degree (Table 1). The study used a five-point Likert scale for the subjectively built environment evaluation of the community, specifically on five levels: very satisfied, relatively satisfied, average, relatively dissatisfied, and very dissatisfied.

3.4.2. Physical and Social Interaction Activities as Mediators

Physical activities and social interaction activities were selected as mediators (Table 1). Regarding physical activities, a lack of regular exercise increases the incidence of common chronic diseases such as hypertension, hyperglycemia, and cardiovascular and cerebrovascular diseases [50,51]. In addition, engaging in physical activities encourages the development of positive emotional states in people, boosts their self-esteem, reduces psychological stress, and lowers the prevalence of depression. Regarding social interaction activities, active social interaction helps older people develop a good sense of community and belonging. A sense of community promotes health-related behaviors and enhances physical and mental health. However, negative social interaction may increase loneliness in older adults, lead to sleep disturbances [52], depression [53], cognitive impairment [54], and even increase mortality. Therefore, the study evaluated the physical activities of older adults in terms of the duration of each leisure physical activity undertaken in the community. We used neighbor interaction degree, neighbor interaction satisfaction, and the number of familiar neighbors as observed variables to reflect the social interaction activities of older people. Neighbor interaction degree and neighborhood interaction satisfaction are also measured on a five-point Likert scale. Neighborhood interaction was categorized as very frequent, frequent, average, seldom, and never. The neighborhood satisfaction was categorized as very satisfied, relatively satisfied, average, relatively dissatisfied, and very dissatisfied.

3.4.3. Physical and Mental Health as Dependent Variables

The self-rated physical and mental health of older adults were used as dependent variables to represent the health level of older adults (Table 2). Although the self-rated health results are subjective assessments of older people, self-rated health includes current and future health expectations and is a multidimensional assessment of one’s health [6,55]. Self-rated health assessment has the advantage that it is easy to obtain a comprehensive picture of an individual’s health status and reflects the actual level of health of older people. A five-point Likert scale measures the physical and mental health of older adults, specifically on five levels: very satisfied, relatively satisfied, average, relatively unsatisfied, and very unsatisfied.

3.4.4. Individual Characteristics as Moderators

Six individual characteristics—length of residence, household registration, local or not, age, gender, and housing property rights—were included as moderators to determine how the relationship between the subjectively built environment and older people’s physical and mental health was moderated by individual characteristics, including the moderation of mediating effects as well as the moderation of direct roles (Table 1).

3.5. Methods

Amos 26.0 was used to construct a dual mediation model to analyze the data. Traditional methods for statistical identification of mediators include the causal method proposed by Baron and Kenny [56], the coefficient difference method proposed by Clogg [57], and the coefficient product method proposed by Sobel [58]. However, the drawbacks of traditional methods of checking mediating effects cannot be ignored [56,57,58]. First, they only perform a simple Z check. Second, mediating roles usually do not conform to the normal distribution. Finally, Z = does not mean that they are necessarily significant. As a result, some scholars have questioned the traditional method. For example, Hayes pointed out that Bootstrap, as proposed by Mackinnon, is a better presence than the traditional method and that there is no need to go back to the traditional method to check for mediated effects [59]. Therefore, this paper uses the Bootstrap technique proposed by Mackinnon for mediating effect checking. The main contents include reestimating the standard error of the indirect effect and the non-standardized coefficient, then calculating the significance level of the indirect effect (Z-value) and estimating the confidence interval of the indirect effect [60].

The specific steps of the analysis path were: first, estimates were conducted on the whole sample to clarify the mechanisms of the subjectively built environment of the community on the physical and mental health of older people. Second, due to China’s unique urban housing system, namely the welfare housing distribution system before 1998 and the security housing system after the market reform, residents are basically in a passive distribution position with little residential self-selection ability [14,38,61]. Therefore, we selected for this study 103 samples of older adults who live in organization-owned housing and indemnificatory apartments and estimated these samples without self-selection of residence. In other words, the study tests the robustness of pathways by which the subjectively built environment of the community affects the physiological and mental health of older adults under the condition of avoiding residential self-selection. Finally, because individual characteristics can act as confounding factors influencing the association between the built environment and health [6], this paper uses individual characteristics of older adults as moderators to test their moderating roles in mediating and direct effects.

4. Results

4.1. Direct Effect of Variables

The results and reference values for the SEM’s fitness metrics are shown in

Table 3. Standard table of the fitting index and analysis results.

Fitting Index	Reference Value	Physical Health	Mental Health
		Model 1	Model 2
Chi-square/DF	1~3	2.492	2.752
CFI	>0.9	0.968	0.958
GFI	>0.9	0.967	0.965
AGFI	>0.9	0.931	0.925
RMSEA	<0.08	0.05	0.054

. All indicators showed that the physical and mental health models fit well with the data. Among them, the potential variable of subjectively built environment included five observed variables: community safety and security, internal supportive living facilities, green environment, beautification degree, and walking environment. Potential variable social interaction activities include three observational variables: neighbor interaction degree, neighbor interaction satisfaction, and the number of familiar neighbors. The impact path coefficients of the SEM reflect the direct impact effect of each variable; the unstandardized path coefficients reflect the significance of the direct impact effect; and the standardized path coefficients reflect the magnitude of the direct impact effect. From the whole sample SEM (Figure 3), each variable’s direct effect is significant and positive.

First, a positive correlation exists between leisure physical and social interaction activities and physical and mental health in older adults. This suggests that increasing leisure physical activities and social interaction activities among older people can effectively promote physical and mental health. Among them, social interaction activities’ influence is superior to leisure physical activities. This may be because, through active social interaction activities, older people can embed themselves in local social networks to alleviate negative emotions such as tension and anxiety in daily life. In addition, through active social interaction activities, they can also build trusting neighborhoods and a safe community environment and increase their willingness to engage in outdoor physical activities, including leisure physical activities. In addition, it is found that the influence of leisure physical activities and social interaction activities on physical health is superior to that on mental health, which may be related to the fact that older adults in China pay less attention to their mental health.

Second, the subjectively community-built environment directly affects the physical and social interaction activities of older adults. Safe, beautiful, walkable, and well-equipped communities can effectively promote leisure physical activities and social interaction activities among older adults. In particular, increasing older people’s satisfaction with a community’s built environment is more conducive to encouraging older adults to participate in social interaction activities than it is to encouraging their participation in leisure physical activities.

Third, the community’s subjectively built environment directly affects older people’s physical and mental health. The higher the level of satisfaction with the built environment, the better the health of older people. In particular, the direct effect of the community’s built environment on mental health is greater than the direct effect on physical health.

Finally, after removing the residential self-selection interference, each variable’s influence direction and significance level are similar to those of the whole sample. However, the significance level of a few paths also changes. The influence strength of each variable also changes significantly (Figure 4). Specifically, there was a significant increase in each variable’s direct effect. The direct effect of the subjectively built environment of the community on health was the only variable showing weakening. The direct effect on physical health was no longer significant, in addition to the distinct role of physical health being directly affected by leisure physical and social interaction activities. These results reflect the fact that the existence of residential self-selection makes older adults who care about their physical and mental health more likely to choose to live in a safe, beautiful, walkable, and humanized space with complete internal facilities. In addition, compared with the whole sample, the impact of leisure physical and social interaction activities on mental health outweighed the influence on physical health. In addition, compared with the whole sample, the influence of leisure physical and social interaction activities on mental health outweighed the impact on physical health. This result is mainly because most non-residential self-selected samples were retired older people living in organization-owned housing. After retirement, leisure time has increased considerably, but social networks have gradually shrunk. There is an urgent need for emotional support and mental comfort in the living environment, for which leisure physical and social interaction activities are vital outlets.

4.2. Mediating Effects of Recreational Physical Activities and Social Interaction Activities

To ensure the robustness of the mediating effects obtained from the SEM, the study performed 5000 Bootstrap sample calculations to get the final model detection results (

Table 4. Mediating effects of social interaction activities and leisure physical activities (whole sample).

Variables	Point Estimate	Product of Coefficients		Bootstrapping				Two-Tailed Significance
				BC 95% CI		Percentile 95% CI
		SE	Z	Lower	Upper	Lower	Upper
Indirect Effects
Subjectively built environment→Social interaction activities→Physical health	0.136	0.036	3.778	0.079	0.221	0.073	0.212	0.00 ***
Subjectively built environment→Leisure physical activities →Physical health	0.018	0.009	2.000	0.005	0.042	0.003	0.038	0.05 **
The difference in mediating effects between social interaction activities and leisure physical activities	0.118	0.037	3.189	0.057	0.203	0.053	0.197	0.00 ***
Total indirect effect	0.154	0.037	4.162	0.095	0.242	0.089	0.232	0.00 ***
Subjectively built environment→Social interaction activities→Mental health	0.047	0.022	2.136	0.011	0.100	0.008	0.094	0.03 **
Subjectively built environment→Leisure physical activities →Mental health	0.011	0.007	1.571	0.001	0.030	0.000	0.027	0.12
The difference in mediating effects between social interaction activities and leisure physical activities	0.036	0.023	1.565	−0.005	0.088	−0.007	0.086	0.12
Total indirect effect	0.058	0.022	2.636	0.021	0.112	0.018	0.106	0.01 ***
Direct Effect
Subjectively built environment→Physical health	0.400	0.072	5.556	0.260	0.544	0.260	0.003	0.00 ***
Subjectively built environment→Mental health	0.422	0.063	6.698	0.302	0.545	0.300	0.544	0.00 ***
Total Effect
Subjectively built environment→Physical health	0.554	0.070	7.914	0.419	0.694	0.418	0.692	0.00 ***
Subjectively built environment→Mental health	0.480	0.063	7.619	0.360	0.604	0.359	0.267	0.00 ***

Note: Physical health is completed in Model 1, while mental health is completed in Model 2. The Point Estimate in the table is a non-standard coefficient. ** and *** represent p < 0.05 and p < 0.001, respectively.

). Firstly, since the study was constructed as a two-mediator model, the study customized the calculation procedure in Amos to calculate the mediating effects of both separately and compare their differences. Secondly, the total and indirect effects were checked to see if they passed the check of Z (Z = unstandardized coefficient/standard error) and if the p-values calculated by the Distribution Calculator were significant. The presence of a significant indirect effect is indicated by the fact that Z ≥ |1.96|, as well as the fact that the reliance intervals for Bias-corrected and Percentile do not contain 0 and have a significant p-value. Finally, the same method was used to determine whether the direct effect was significant. If significant, the mediating variable has a partial mediating effect; if not, it has a full mediating effect. The study found:

First, leisure physical and social interaction activities play a significant partial mediating role between the community’s subjectively built environment and older people’s physical health. The mediating effect of social interaction activities is superior to the mediating role of leisure physical activities. Specifically, the total effect of the subjectively built environment of the community on the physical health of older adults is significant (0.554); the total indirect effect accounts for about 27.8% of the total effect, of which 24.5% comes from the mediating role of social interaction activities, and 3.3% comes from the intermediary role of leisure physical activities. Therefore, the mediating effect of social interaction activities is superior to that of physical leisure activities. Since the subjectively built environment of the community has a significant direct impact on physical health, leisure physical activities and social interaction activities play a partial mediating role between the subjectively built environment of the community and the physical health of older adults. It can be seen that the subjectively built environment of the community plays a positive guiding role in the physical health of older adults through leisure physical activities and social interaction activities. Therefore, increasing the satisfaction of older adults with the community’s built environment can effectively increase the amount of leisure physical activities and improve the satisfaction and quality of social interaction activities, thus indirectly improving the physical health of older adults.

Second, social interaction activities play a significant, partially mediating role between the subjectively built environment of the community and the mental health of older adults, while the mediating effect of leisure physical activities is not significant. After calculation, the total effect of the subjectively built environment of the community on the mental health of older adults is 0.480. The effect of influencing mental health through mediators accounts for 12.1% of the total effect; 11.1% of the influence effect comes from the mediating role of social interaction activities; the mediating role of leisure physical activities is not significant. The subjectively built environment of the community has a significant direct effect on the mental health of older adults, so social interaction activities play a partial mediating role. It can be seen that the subjectively built environment of the community plays a positive guiding role in the mental health of older adults through social communication activities. The built environment guidance promotes the investment in social interaction activities to a certain extent and then converts into the return of mental health.

To sum up, the changes in social interaction activities and leisure physical activities of older adults are, to a certain extent, the product of the choice of the surrounding environment. The environment-based social interaction activities and leisure physical activity intervention measures cover a wider range, which may affect the behavior choices of the entire older adults’ group. Therefore, the built environment plays a crucial role in the health promotion of older adults. Optimizing the community’s built environment is a feasible way to improve the physical and mental health of older adults. It is necessary to consider the leisure and social environment in planning.

The calculation results of the sample without residential self-selection (

Table 5. Mediating effects of social interaction activities and recreational leisure activities (no self-selection sample).

Variables	Point Estimate	Product of Coefficients		Bootstrapping				Two-Tailed Significance
				BC 95% CI		Percentile 95% CI
		SE	Z	Lower	Upper	Lower	Upper
Indirect Effects
Subjectively built environment→Social interaction activities→Physical health	0.209	0.052	4.019	0.032	0.244	0.027	0.229	0.00 ***
Subjectively built environment→Leisure physical activities →Physical health	0.105	0.099	1.061	0.058	0.447	0.056	0.441	0.29
The difference in mediating effects between social interaction activities and leisure physical activities	−0.103	0.114	−0.904	−0.358	0.089	−0.351	0.092	1.63
Total indirect effect	0.314	0.11	2.855	0.142	0.58	0.138	0.572	0.00 ***
Subjectively built environment→Social interaction activities-->Mental health	0.172	0.099	1.737	0.02	0.433	−0.003	0.388	0.08 *
Subjectively built environment→Leisure physical activities →Mental health	0.099	0.051	1.941	0.02	0.229	0.012	0.215	0.05 **
The difference in mediating effects between social interaction activities and leisure physical activities	−0.073	0.119	−0.613	−0.345	0.131	−0.316	0.154	1.46
Total indirect effect	0.270	0.104	2.596	0.101	0.522	0.080	0.489	0.01 ***
Direct Effect
Subjectively built environment→Physical health	0.251	0.170	1.476	−0.092	0.576	−0.119	0.556	0.14
Subjectively built environment→Mental health	0.249	0.148	1.682	−0.060	0.518	−0.059	0.518	0.09 *
Total Effect
Subjectively built environment→Physical health	0.565	0.141	4.007	0.277	0.829	0.274	0.825	0.00 ***
Subjectively built environment→Mental health	0.519	0.135	3.844	0.244	0.782	0.227	0.768	0.00 ***

Note: Physical health is completed in Model 1, while mental health is completed in Model 2; The Point Estimate in the table is a non-standard coefficient. *, ** and *** represent p < 0.01, p < 0.05 and p < 0.001, respectively.

) suggest the total effect of the subjectively built environment of the community on the physical and mental health of older adults was 0.565 and 0.519, respectively. The total indirect effect proportion increased to 55.6% and 52.3%, respectively. This result is related to the fact that the direct effect of the subjectively built environment of the community on the physical health of older adults is no longer significant, and the direct effect on the mental health of older people is weaker. Among them, in physical health, 37% of the indirect effects originate from the mediating role of social interaction activities, while the mediating effect of leisure physical activities is not significant. In mental health, 33.1% and 19.2% of the indirect effects originate from the mediating role of social interaction activities and leisure physical activities, respectively.

The comparison of the non-resident self-selection sample’s action path with the whole sample’s action path suggests that after removing the interference of the resident self-selection factor, the relationship between each variable is generally consistent, indicating that the above conclusions are generally stable. However, there are some differences. First, the indirect effect influence is significantly enhanced. By contrast, the direct effect’s influence is weaker. Even social interaction activities play a complete mediating role between the community’s subjectively built environment and the physical health of older adults. Second, the mediating effect of leisure physical activities is no longer significant between the community’s subjectively built environment and physical health. However, it plays a partial mediating role between the community’s subjectively built environment and mental health, which is the opposite of the conclusion from the whole sample. Finally, social interaction and leisure physical activities play a relatively balanced mediating role between the community’s subjectively built environment and mental health.

4.3. Moderating Effects of Individual Characteristics

This study conducted a separate test on the moderating effect of length of residence, household registration, whether people are local, gender, age, and housing property rights. Furthermore, the moderating effects of these individual characteristics include the moderation of mediating effects and the moderation of direct effects.

Table 6. Mediated moderating role of individual attributes (physical health).

Physical Health		Leisure Physical Activities				Social Interaction Activities
Moderator	Level	Conditional Indirect Effect	SE	Z	p	Conditional Indirect Effect	SE	Z	p
Length of residence	Less than 5 years	−0.005	0.016	−0.313	−	0.144	0.106	1.358	−
	More than 5 years	0.032	0.014	2.286	0.02 **	0.122	0.037	3.297	0.001 ***
Household registration	Urban	0.016	0.013	1.231	−	0.180	0.05	3.6	0.00 ***
	Rural	−0.031	0.032	−0.969	−	−0.120	0.122	−0.984	−
Local or not	Yes	0.035	0.014	2.5	0.001 ***	0.138	0.04	3.45	0.001 ***
	No	0.001	0.044	0.023	−	0.134	0.127	1.055	−
Gender	Male	0.027	0.022	1.227	−	0.071	0.037	1.919	−
	Female	0.008	0.01	0.8	−	0.225	0.07	3.214	0.001 ***
Age	Under 65 years old	0.006	0.014	0.429	−	0.06	0.068	0.882	−
	65 years and over	0.027	0.012	2.25	0.01 ***	0.154	0.044	3.5	0.00 ***
Housing property rights	No self-selection	0.005	0.007	0.714	−	0.118	0.038	3.105	0.00 ***
	Have self-selection	0.105	0.052	2.019	0.04 **	0.209	0.096	2.177	0.03 **

Note: Completed in Model 1. The Point Estimate in the table is a non-standard coefficient. ** and *** represent p < 0.05 and p < 0.001, respectively.

and

Table 7. Mediated moderating role of individual attributes (mental health).

Mental Health		Leisure Physical Activities				Social Interaction Activities
Moderator	Level	Conditional Indirect Effect	SE	Z	p	Conditional Indirect Effect	SE	Z	p
Length of residence	Less than 5 years	0.002	0.011	0.182	−	0.086	0.053	1.623	−
	More than 5 years	0.017	0.01	1.7	−	0.036	0.022	1.636	−
Household registration	Urban	0.009	0.009	1	−	0.082	0.029	2.828	0.005 **
	Rural	0.001	0.025	0.04	−	−0.102	0.063	−1.619	−
Local or not	Yes	0.017	0.01	1.7	−	0.056	0.027	2.074	0.03 **
	No	−0.002	0.026	−0.077	−	0.018	0.046	0.391	−
Gender	Male	0.027	0.016	1.688	−	0.015	0.017	0.882	−
	Female	0.003	0.006	0.5	−	0.108	0.048	2.25	0.02 **
Age	Under 65 years old	0.001	0.01	0.100	−	0.026	0.037	0.703	−
	65 years and over	0.018	0.01	1.800	−	0.054	0.026	2.077	0.04 **
Housing property right	No self-selection	0.001	0.005	0.200	−	0.020	0.020	1	−
	Have self-selection	0.099	0.051	1.941	−	0.172	0.100	1.72	−

Note: Completed in Model 2. The Point Estimate in the table is a non-standard coefficient. ** and *** represent p < 0.05 and p < 0.001, respectively.

show the estimates, standard errors, z-statistics, and significance values of the conditional indirect effects for the mediated effects of social interaction activities and leisure physical activities moderated by the six attribute characteristics of older adults [62].

From the perspective of the physical health effects of the community’s subjectively built environment (Table 6), first, the length of residence, household registration, whether people are local, gender, and age have a moderating effect on the intermediary effect of social interaction activities. Specifically, when older adults are urban registered residence women over 65 years old and have lived in the community for more than 5 years, the mediating effect of social interaction activities is significant. However, when older adults are male outsiders with rural registered residences under 65 years old and have lived in the community for less than 5 years, the mediating effect of social interaction activities is not significant. Second, housing property rights (residential self-selection) have no moderating effect on the mediating role of social interaction activities. The latter is significant under residential self-selection and non-residential self-selection, and there is no significant difference between the two. Third, length of residence, whether the residents are locals, their age, and housing property rights (residential self-selection) have a moderating role in the mediating effect of leisure physical activities. Specifically, when older adults are local people who have lived in the community for more than 5 years and have residential self-selection, the mediating effect of leisure physical activities is significant. When older adults are outsiders living in the community for less than 5 years under the age of 65 and have no residential self-selection, the mediating effect of leisure physical activities is insignificant. Fourth, household registration and gender of older adults have no moderating role in the mediating effect of leisure physical activities.

To sum up, length of residence, whether people are local, and age all moderate the mediating effects of social interaction and leisure physical activities. Household registration and gender only have moderating roles in the mediating effect of social interaction activities. In comparison, housing property rights (residential self-selection) only have a moderating role in the mediating effect of leisure physical activities.

From the perspective of the mental health effect of the subjectively built environment of the community (Table 7), first, the six individual characteristics have no moderating role in the mediating effect of leisure physical activities, which is mainly related to the non-significant mediating effect of leisure physical activities. Second, the length of residence and housing property rights (residential self-selection) have no moderating role in the mediating effect of social interaction activities. Finally, registered residence, gender, age, and whether people are local have a moderating role in the mediating effect of social interaction activities. Specifically, the mediating effect of social interaction activities is significant when older adults are local urban women aged 65 years and above but not significant when older adults are rural men aged 65 years and below.

In addition, the moderating role of six individual characteristics in the direct impact of the community’s subjectively built environment on physical and mental health was tested (

Table 8. Moderating effect of individual attribute characteristics on direct effects.

Variables	Moderator	Confounding Effect		Level	Unstandardized	Standardized
		CMIN	p		Coefficients	Coefficients
Subjectively built environment→Physical health	Household registration	6.98	0.008 ***	urban	0.316 ***	0.198
				rural	0.836 ***	0.557

Note: Completed in Model 1. The Point Estimate in the table is a non-standard coefficient. *** represent p < 0.001, respectively.

). Only in the physical health effects of the community’s subjectively built environment does household registration significantly moderate the direct impact (p = 0.008). Specifically, when older adults are rural residents, the subjectively built environment of the community has a superior direct impact on physical health (standardization coefficient = 0.557), but the direct impact is weak when older adults are urban residents (standardization coefficient = 0.198).

4.4. The Impact Mechanism of the Community’s Subjectively Built Environment on the Physical and Mental Health of Older Adults

The community’s built environment does not directly affect the physical and mental health of older adults. It often affects the health of older adults by changing their health-related behaviors. Therefore, this paper discusses the mechanism of the “community-built environment—the physical and mental health of older adults” with leisure physical and social interaction activities as mediators. The action mechanisms are divided into “community-built environment—leisure physical activities—physical health” and “community-built environment—social interaction—physical and mental health.”

The first is the “community-built environment—leisure physical activities—physical health” mechanism. Older people face various health problems, and appropriate physical activities can improve their health and prevent the onset of related chronic diseases. The community-built environment is a vital factor that affects the leisure physical activities of older adults. Research shows that the duration of leisure physical activities among older adults is significantly positively related to the subjectively built environment of the community. Older adults with high satisfaction with community safety and security, a green environment, a walking environment, a beautification degree, and internal supportive living facilities have a higher level of leisure physical activities. In addition, our study demonstrated that leisure physical activities partially mediate the relationship between the subjectively built environment of the community and the physical health of older adults. Therefore, a well-built community environment can promote the physical health of older adults by strengthening their leisurely physical activities.

The second is the “community-built environment—social interaction activities—physical and mental health” mechanism. It was found that compared with leisure physical activities, social interaction activities not only mediate the impact of the community-built environment on the physical and mental health of older adults but also mediate effects more strongly. In addition, the positive impact of the community’s subjectively built environment on the social interaction activities of older adults and the positive influence of social interaction activities on the physical and mental health of older adults are stronger than leisure physical activities. Social communication activities play a crucial role in the community-built environment’s physical and mental health effects. The active social communication activities of older adults can promote health-related behaviors, including leisure physical activities. It can also promote the rapid diffusion of health-related information in the community, broaden the social participation channels of older adults, and reduce depression, anxiety, loneliness, and other negative emotions. The reason for this is that frequent social interactions can expand social networks for older adults, accumulate social capital, and enhance the sense of connection, belonging, and cohesion of older adult communities. Older adults are nostalgic for a place because they are nostalgic for the people, services, and interpersonal relationships there. Without these attributes, the community is just a meaningless space for older adults, making them unwilling to go out for activities or even leading to complete isolation. Social interaction activities become an opportunity for older adults to build a “sense of place” in the community, turn it into a friendly place, and stimulate older adults to go out. A high-quality community-built environment can improve the health of older adults by promoting social interaction activities.

5. Discussion

Using data from a questionnaire survey conducted in Hefei in 2021, this study applied SEM to explore how elements of the subjectively built environment in cities affect the physical and mental health of older adults in a combination of ways and to uncover the mechanisms of influence. The findings were also tested for robustness by excluding residential self-selection. To the best of our knowledge, few studies have discussed the impact of subjectively built environmental elements on health in an integrated manner and revealed their pathways of influence. The findings suggest that not only do the elements of the subjectively built environment of the community in an integrated manner influence the physical and mental health of older people through mediating variables, but also residential self-selection partially influences the findings.

The subjectively built environment has a significant impact on the physical and mental health of older adults, consistent with previous findings [6,47,63,64]. For example, a study in Columbus, USA, found that perceived pedestrian-friendly neighborhood characteristics were significantly and positively associated with self-rated health [64]. A study in China also observed that the perceived urban built environment significantly impacted residents’ self-rated health, such as their perceptions of urban greenery, infrastructure conditions, housing security, and health services [6]. Existing studies have only explored the independent effects of individually built environmental elements on health. However, our study reveals the impact of specific combinations of subjectively built environmental elements (such as community safety and security, internal supportive living facilities, the green environment, the walking environment, and the degree of beautification) on older people’s physical and mental health. Therefore, this study indicates that the planning and construction of urban built environments should not only pay attention to a single aspect of the built environment, but should also consider all elements of the built environment in a comprehensive way, as the single element of the built environment focused on may have no impact on the physical and mental health of older adults, but the combination of other built environment elements will have an impact on them.

The community-built environment does not directly affect the physical and mental health of older adults. It can affect the health of older adults by changing their leisure physical activities and social interaction activities. By contrast, the explanatory effect of leisure physical activity as a mediating variable is generally smaller than that of social interaction activity. This result is similar to previous findings [14,18]. Communities with environmental features such as good walkability, good amenities, ample open space, and pleasing environmental design have residents who spend more time outdoors and receive more health benefits [65]. However, the existing literature has not considered how the differences in the mechanisms of the built environment affect the mental and physical health of older adults. This study found that physical and mental health mechanisms are different. Two functional mechanisms, “community-built environment—leisure physical activities physical health” and “community-built environment—social interaction activities—physical and mental health,” were defined. Therefore, planners need to determine the focus of the built environment design, deciding whether to focus on physical health or mental health. If the focus is only on physical health, a built environment planning strategy should be developed to guide older adults into participating in physical activities. If the focus is on mental health, the social interaction needs of older adults also need to be considered.

After avoiding the interference of residential self-selection, the above research conclusions are generally stable, but a few impact paths have changed. This is supported by some literature [1,66]. However, the changes that occur after avoiding residential self-selection still remind us that residential self-selection has a specific interfering effect on the health effects of the built environment. Therefore, it needs to be considered in research, such as that by Durand et al. [24] and Zhang et al. [15], who control the interference of residential self-selection through the propensity value matching method. Most studies in China use demolition and resettlement housing to strip away the impact of residential self-selection [41]. After controlling for the self-selection effect of housing, the main changes found in this study are as follows: first, the mediating effect of social interaction activities and leisure physical activities is significantly enhanced; second, the path of the physical and mental health effects of the subjectively built environment in the community is exactly opposite to that of the whole sample. The mediating effect of leisure physical activities on physical health is no longer significant but plays a significant mediating role in mental health; finally, leisure physical activities and social interaction activities play a relatively balanced mediating role in the mental health effects of the subjectively built environment of the community.

The individual characteristics of older adults have been observed to have a significant moderating effect. This result is inconsistent with most current studies that control individual characteristics as covariates [45]. However, the moderating effect of individual characteristics is also supported by the literature [46]. Our findings show that among the physical health effects of the community’s subjectively built environment, length of residence, whether people are local, and age all play a moderating role in the mediating effect of social interaction activities and leisure physical activities. Household registration and gender only have a moderating role in the mediating effect of social interaction activities. In comparison, housing property rights (residential self-selection) only have a moderating role in the mediating effect of leisure physical activities. Among the mental health effects of the community’s subjectively built environment, household registration, gender, age, and whether people are local have moderating effects on social interaction activities. By contrast, residence years and housing property rights (residential self-selection) have no moderating effects on social interaction activities. In addition, only the direct impact of the community’s subjectively built environment on physical health is significantly moderated by household registration. Therefore, planning scholars and policymakers need to consider the heterogeneity of the impact of the urban built environment on the physical and mental health of older adults in different social groups, and should formulate targeted living circles for different groups according to local conditions, paying particular attention to socially vulnerable groups and promoting environmental equity and justice.

6. Conclusions

This study used a SEM to examine the pathways through which the subjectively built environment of the community affects older people’s physical and mental health. The specific combination of the community’s subjectively built environmental elements influences older people’s physical and mental health through leisure physical and social interaction activities. The pathways of influence differ between physical and mental health. On the one hand, leisure physical and social interaction activities play a mediating role in the influence of the community’s subjectively built environment on the physical health of older people. On the other hand, social interaction activities alone play a mediating role in the influence of the community’s subjectively built environment on the mental health of older people.

By analyzing the impact and mechanism of the community’s subjectively built environment on the physical and mental health of older adults, we can explore the path to promoting the physical and mental health benefits of older adults from the perspective of environmental perception. We propose the operational direction and strategies that can be implemented to achieve healthy aging and healthy urban construction. As older adults are limited by physical and cognitive functions and other reasons, their range of travel is limited, and the community becomes the main place for daily activities. The research results also show that older adults are more sensitive to the community-built environment, such as internal supportive living facilities, green environments, walking environments, beautification degrees, and community safety and security. Optimizing the community-built environment is an effective way to improve the physical and mental health of older adults. Improving the types of community support leisure, landscape, and exercise facilities, beautifying the community environment, and enhancing community safety can promote leisure physical activities and social interaction activities among older people, thereby reducing the incidence of chronic diseases in older people, increasing their sense of community identity, belonging, and security, and thus enhancing their physical and mental health.

In terms of the selection of built environment indicators, this study uses community subjectively built environment indicators, which provide more environmental, individual behavioral, and psychological information than objectively measured built environment indicators. In the future, however, it will be possible to incorporate both the subjectively and objectively built environment into a comprehensive theoretical framework and to compare their relative importance. In addition, a single micro-indicator (such as the number of benches or the distance between buildings) is separated from the macro-scale built environment (such as the degree of land use mixing) to verify its relationship with the physical and mental health of older adults, and it is easy to get an accurate conclusion. The change in the macro-scale built environment requires a large amount of capital and takes a long time, so it is difficult to have a substantial change in a short period of time. By contrast, the micro-elements are highly changeable and easy to update, so they have strong realizability. Therefore, we can try to adopt micro-scale objectively built environment indicators in the future.

There are still limitations to this paper. First, the indexes of leisure physical activities and the physical and mental health of older adults are based on the retrospective survey. More objective and accurate measurement methods, such as a three-dimensional accelerometer, disease statistics, and mental health status assessment, can be adopted in the future. Second, due to the lack of multidimensional panel data, the conclusion relies on cross-sectional data, which makes it challenging to understand the dynamic relationship between changes in the built environment and health. Longitudinal studies utilizing follow-up investigations or quasi-experiments are needed to confirm the causal relationship between the built environment and the physical and mental health of older adults.