Associations between regular physical exercise and physical, emotional, and cognitive health of older adults in China: an 8-year longitudinal study with propensity score matching

Background The importance of healthy aging is growing in China as it has the largest number of older adults in the world and is one of the fastest-aging countries. This study aimed to examine the predictive value of regular physical exercise in relation to the physical, emotional, and cognitive health among samples of adults aged ≥60 years in China during an 8-year period. Methods A total of 10,691 older adults were extracted from two waves of national data from the China Family Panel Studies in 2010 and 2018. To minimize the impact of selection bias on the findings, a longitudinal propensity score matching (LPSM) method was used to examine the relationships between regular physical exercise and emotional health (depression), between regular physical exercise and physical health (instrumental activities of daily living), and between regular physical exercise and cognitive health (cognitive ability) of older adults. After LPSM, 856 older adults were included in the study. In the regular physical exercise group, the average age of participants at baseline year was 65.67 years, with an average age of 65.90 years for 238 men and 65.45 years for 190 women, and in the non-physical exercise group, their average age at baseline year was 65.70 years, with an average age of 65.45 years for 253 men and 65.98 years for 175 women. Results LPSM indicated that regular physical exercise has been found to be effective in improving physical function and reducing depressive symptoms in old adults, even after controlling for background differences. However, the sensitivity analysis suggests that the positive association between regular physical exercise and cognitive function may not be sufficiently valid. Conclusion The findings of this study indicate that engaging in long-term structured and repetitive physical exercise can have a significant positive effect on reducing depressive symptoms and improving the physical function of older adults. As a result, incorporating regular physical exercise into the lifestyle of older adults is recognized as an effective strategy for promoting healthy aging and reducing the strain on public health resources.


Introduction
Population aging is considered a significant achievement for humanity, but it also presents significant challenges.China, with the largest number of older adults in the world, is one of the fastest-aging countries (1).By 2021, China had approximately 267 million people aged 60 years and above.However, like many other nations facing the fast-growing aging population for the first time in history, the gap in the literature regarding the promotion of sustainable good healthrelated quality of life in aging populations is hindering the advancement of China toward healthy aging, an important global issue that warrants immediate investigation (2).
Aging is a progressive process that leads to the deterioration of various physiological functions in organisms (3).One of the significant declines observed from midlife to old age is the gradual decrease in physical health (4)(5)(6).This decline can result in the loss of physical independence, an increased risk of falls, and a diminished quality of life (7)(8)(9).Mental health problems are also common among older adults (10,11).It is estimated that at least one in four older adults experience mental disorders such as depression, anxiety, or dementia, and this number is expected to double by 2030 (12)(13)(14).In addition, cognitive decline is a common health issue among older adults.According to a CDC report in the United States, adults aged over 64 years have a higher prevalence of subjective cognitive decline (11.7%) than adults aged 45-64 years (10.8%)(15).A systematic review published in 2021 found that the prevalence of mild cognitive impairment (MCI) among Chinese community-dwelling populations over 55 years old was 12.2%, and it increased with age (16).In conclusion, there is a pressing need to identify effective strategies that can support healthy aging and maintain or enhance physical, emotional, and cognitive health in older adults.
Regular physical exercise is a major contributor to healthy aging, offering a low-cost, low-risk, and effective strategy for addressing deteriorating health conditions in older adults (17)(18)(19).Numerous epidemiological studies have consistently demonstrated the positive effects of regular physical exercise on the human body.These effects include reducing frailty, improving muscle strength and endurance, lowering body mass index (BMI), and decreasing the risk of major mortality factors.Regular physical exercise has also been found to extend life expectancy in patients with arterial hypertension (20), diabetes mellitus type 2 (21), dyslipidemia (22), coronary heart disease (23), stroke (24), or cancer (25).Additionally, research consistently indicates that regular physical exercise has positive effects on emotional health (26,27), with a particular focus on reducing depressive symptoms (28,29) and preventing the occurrence of depression (30,31).Exercise plays a role in regulating neurotransmitter levels, such as serotonin and dopamine, which are essential for mood regulation (32).It also promotes neurogenesis and neuroplasticity, enhancing mood stability (33).In addition, exercise has antiinflammatory effects and reduces stress hormones, which contribute to alleviating depression (34).The release of endorphins during exercise acts as a natural mood elevator.Furthermore, cognitive distraction and social interaction during physical activity also play roles in improving mental wellbeing (35).Overall, scientific evidence highlights that exercise provides a holistic approach to mitigating depressive symptoms and enhancing mental health.Inconsistent findings also have been reported regarding the associations between regular physical exercise and cognitive functions (36).However, some studies have shown a positive correlation between regular physical exercise and the maintenance or enhancement of cognitive function (37,38).Other studies have reported a dose-response relationship (39-41) or found no significant correlation (42).
While there are numerous benefits of regular physical exercise on the emotional, physical, and cognitive health of older adults, several important challenges still persist.
The first significant challenge is that many studies examining the impact of regular physical exercise on emotional, physical, and cognitive health in older adults have small sample sizes, which limits their ability to demonstrate strong effects.For instance, a systematic review of 14 randomized controlled trials focused on the impact of exercise on cognitive outcomes in adults with mild cognitive impairment.The review revealed that most of the trials had small sample sizes (n < 200), which restricted their ability to detect small effects.Only 8% of the cognitive outcomes were statistically significant.These findings suggest a lack of strong and consistent evidence supporting significant and robust cognitive improvement through exercise in individuals at risk of dementia (43).Second, regular physical exercise recommendations for health are generally based on studies with a cross-sectional design or a short-term longitudinal design using young or middle-aged adults or clinical populations.Finally, selection bias is a significant concern in studies on regular physical exercise.It is widely recognized as one of the most substantial biases that can impact the findings of empirical studies (44).The existing literature suggests that various demographic and socioeconomic variables are associated with the health of older adults (45,46).These variables include gender, age, education level, marital status, household registration, income, interpersonal relationships, chronicity, and the number of children.For instance, men and women exhibit physiological and cultural differences that contribute to disparities in health outcomes (46,47).Age also plays a significant role, as the immune system weakens with age, resulting in a decline in both physical and mental health (45,46).Higher education levels and being married are protective factors for health.Moreover, urban older adults in China generally experience better health than their rural counterparts (48)(49)(50)(51).Having a higher income or good interpersonal relationships also promotes better health (52,53).Chronic illnesses can directly harm an individual's physical health and lead to negative emotions (54).Finally, the number of children can influence the economic and social behavior of parents, which subsequently affects the health of older adults (55).Therefore, these natural differences between older adults who do regular physical exercise and those who do not will affect the external validity of the analysis and lead to unreliable research outcomes.
To address these challenges, the present study aimed to further examine the predictive value of regular physical exercise in relation to physical, emotional, and cognitive health.To address confounding factors and selection bias, we conducted a longitudinal propensity score matching analysis (LPSM).The selection of suitable confounding variables is the basis for applying LPSM.Based on the literature review, we identified more than 10 confounding variables between regular physical exercise and physical, emotional, and cognitive health in older adults.To generate national results, we focused on older adults aged ≥60 years using data from the China Family Panel Studies (CFPS), which is a comprehensive, longitudinal social survey covering a wide range of social phenomena in contemporary China (56).Finally, we analyzed the long-term longitudinal effect of regular physical exercise using matched data from the 2010-2018 waves of CFPS.
This study utilized CFPS data to assess the impact of regular physical exercise on the physical, emotional, and cognitive health of older adults aged ≥60 years in China over an 8-year period.The longitudinal propensity score matching analysis (LPSM) method was employed to minimize the influence of potential confounding factors.

Participants
The data used for this study was gathered from the CFPS of the years 2010 and 2018.The CFPS is a longitudinal research project administered by the Institute of Social Science Survey (ISSS) at Peking University, China.A multi-stage probability-proportional-to-size (PPS) strategy with implicit stratification was performed in the sampling process that comprises three stages: the county level as the primary sampling unit, a community or village for the second-stage sampling unit, and household as the final sampling unit.CFPS surveyed respondents in sampling units in 25 provinces and consisted of a rich set of socioeconomic questions and information on the levels of individual, family, and community.Details of the original study have been reported elsewhere (56,57).The Peking University Biomedical Ethics Review Committee provided ethical approval of the survey (approval number: IRB00001052-14,010). Respondents were given a statement explaining the purpose of the study, and all study participants signed a written informed consent prior to being investigated.All methods will be carried out in accordance with relevant guidelines and regulations of the Declaration of Helsinki.
Following an initial baseline survey wave in 2010, ISSS conducted four follow-up survey waves in 2012, 2014, 2016, and 2018.For the purposes of the survey, first, we selected 7,040 older adults who were at or over 60 years of age from the 2010 data of CFPS and 3,651 older adults who were 68 years old or older from the 2018 data of CFPS.In total, 3,170 subjects remained by matching ID in 2010 and 2018, and subjects with missing data on relevant variables in this study were deleted.As a result, 1,927 subjects were left.Second, among them, 761 older adults were with regular physical exercise, 1,030 older adults were without physical exercise, and 135 older adults with non-regular physical exercise were not included in this study.Third, by the method of LPSM, 761 older adults in the regular physical exercise group and 1,031 older adults in the non-physical exercise group were matched on 12 covariates, and 428 subjects remained in each group finally (see Figure 1 for the sample selection).

Eligibility criteria
Participants meeting the following inclusion criteria were included in this study: (a) individuals aged 60 years or above in 2010; (b) individuals who participated in both the 2010 and 2018 waves; and (c) individuals with no missing data on relevant variables.In the 2010 CFPS data, there were 7,040 older adults aged 60 years and above, who would be 68 years and above in 2018.In the 2018 CFPS data, there were 3,650 older adults aged 68 years and above, of which 3,170 had also participated in the 2010 CFPS.After excluding cases with missing data in any relevant variable, a total of 1,927 subjects remained.Among them, 761 older adults were in the regular physical exercise group, 1,031 older adults were in the non-physical exercise group, and 135 older adults with non-regular physical exercise were not included in this study.Using the LPSM method, 761 older adults in the regular physical exercise group and 1,031 older adults in the non-physical activity group were matched on 12 covariates, resulting in 428 subjects remaining in each group (see Figure 1 for the sample selection).

Physical health
The physical health of older adults was evaluated by the instrumental activities of daily living (IADLs), which were confirmed to be a core set of essential activities for individuals to live independently and the key measures of physical health for older adults in the existing studies (58)(59)(60) IADLs in CFPS were measured by a modified version of the Lawton IADLs Scale (61).Respondents were asked whether they could perform these seven tasks without assistance: (a) outdoor activities (e.g., walking 300 m); (b) kitchen chores (e.g., preparing meals and washing dishes); (c) shopping for groceries; (d) having meals; (e) managing transportation; (f) housekeeping (cleaning); and (g) doing laundry.A binary variable was constructed for IADLs, with 0 representing difficulty with any of the seven IADLs and 1 representing "no" difficulty with any of the seven IADLs.Previous research has demonstrated that the Chinese version of the IADLs scale exhibits strong internal consistency (coefficient alpha = 0.86) and good test-retest reliability (r = 0.90) (62).The IADLs Scale in CFPS has proven to be a valid measure of the daily performance of older adults (63,64).Measurement of IADLs functions is essential as it serves as an early sign of functional decline in old age.These measurements are also predictors of the need for alternative living arrangements, the utilization of paid home care, and admission to nursing homes (64).In the current sample, the IADLs Scale shows good internal consistency (coefficient alpha = 0.83).

Cognitive health
The cognitive health of older adults was evaluated by cognitive ability tests in CFPS (56).The theoretical basis of the CFPS cognitive ability tests was the design of the Guttman Scale in psychometry, which showed good reliability and validity (66).Cognitive ability tests in CFPS applied both vocabulary and mathematics tests as measurement tools to measure and represent "crystallized intelligence, " which is referred to as acquired knowledge through learning, experience, and education (56,67).Previous research has shown that the vocabulary and mathematics tests in CFPS have strong internal consistency (coefficient alpha = 0.85-0.96)and criterion validity (r = 0.80-0.86)(68).These tests have been widely used and validated for measuring cognitive ability in Chinese adults and adolescents in previous studies (66,(69)(70)(71)(72).
The vocabulary test consists of 34 word-recognition questions, which measure one's vocabulary by how difficult of a character he or she can recognize and are sorted in ascending order of difficulty.The respondents were asked to recognize the increasingly difficult characters one by one until they failed to recognize three consecutive characters.The final test score was based on the rank order of the last character recognized by each respondent, ranging from 0 (lowest) to 34 (highest).
The mathematics test consists of 24 mathematical questions.The procedures for the mathematics test were similar to those mentioned above.The test score was assigned using the same rank-order rule as that in the vocabulary test and recorded on a scale from 0 (lowest) to 24 (highest).and (f) I felt that life could not continue; and two questions measure positive mood: (g) I was happy and (h) I enjoyed life.These items have four answer options: almost none (less than 1 day), sometimes (1-2 days), often (3-4 days), and most of the time (5-7 days).The corresponding score of each option is 1, 2, 3, and 4. The higher the total score is, the more severely depressed the subjects feel.CESD-8 is a widely used measure for screening depression in large-scale surveys among older people (74).It offers several advantages, including brevity, accessibility, and ease of use by non-mental health professionals in non-clinical settings.Additionally, it has demonstrated reliability and validity compared to longer scales (74).In the Chinese context, the CESD-8 has been extensively utilized and validated as a valuable and dependable tool for identifying the risk of depression among the older population, aiding in further diagnosis (75-77).Cronbach's alpha of the study for the CESD-8 was found to be 0.85 and 0.81 for the 2010 and 2018 data, respectively.

Regular physical exercise
According to the previous study (78)(79)(80), regular physical exercise was defined as planned, structured, and repetitive physical activity with the ultimate or intermediate goal of improving or maintaining physical fitness, and the frequency and duration of exercise were usually chosen to reflect the physical activity status of the population.
In the CFPS, physical exercise includes walking, long-distance running, jogging, and hiking; martial arts such as taijiquan and qigong exercises; and indoor and outdoor activities such as dancing, aerobics, gymnastics, and yoga, as well as ball sports with small and large balls, water sports such as swimming, diving, boating, and sailing, winter ice and snow sports, and contact sports such as wrestling, judo, and boxing.The following two questions were used in the CFPS to determine whether older adults do regular physical exercise ( 56): (a) Regarding the frequency of physical exercise, respondents were asked, "How often do you participate in physical exercise?"Responses were categorized into the following seven levels: never; less than 1 time per month on average; more than 1 time per month on average, but less than 1 time per week; 1-2 times per week on average; 3-4 times per week on average; 5 or more times per week on average; and 1 time a day; and (b) regarding the duration of physical exercise, respondents were asked, "In the past 12 months, how many minutes did you exercise at a time?"As such, this variable is also a continuous variable.With reference to previous studies on the division of regular physical exercise and the answers to the above two questions, we divided the participants into two groups: (a) regular physical exercise group (participants engaged in physical exercise "at least three times a week" and at least 30 min per time in two waves of CFPS surveys in 2010 and 2018) and (b) non-physical exercise group (participants who indicated that they "never" do any physical exercise in two waves of CFPS surveys in 2010 and 2018).

Covariates
Based on the previous studies (46,47,51,81,82) and information in the 2010 and 2018 CFPS, this study initially identified 12 potential confounding covariates: gender, age, number of children, educational status, household registration, marital status, income level, chronic diseases, interpersonal relationships, physical (2010), depression (2010), and cognitive (2010) health.Educational status included the following: 1-illiteracy, 2-elementary school, 3-junior high school, 4-high school/vocational high school, 5-junior college, 6university undergraduate, 7-master, and 8-doctorate; household registration included the following: 0-rural and 1-urban; marital status included the following: 0-having no spouse and 1-having a spouse; income level was measured by one question: How do you perceive your income level in local locations?(1-very low, 5very high); chronic disease included the following: 0-having no chronic disease and 1-having at least one chronic disease; and interpersonal relationships were measured by one question: How do you perceive your interpersonal relationships?(0-worst; 10-best).

Statistical analysis
In this study, LPSM is the key statistical technique used to effectively eliminate sample-selection bias and mixed bias by matching the individuals in the treatment group with the comparable objects in the control group (83,84).LPSM analyses are generally divided into two steps (83): The first analysis is used to calculate the propensity score and match subjects according to the propensity score.The propensity score estimated by the logit regression model was employed to match each subject between the regular physical exercise and non-physical exercise groups in similar conditions, and it is estimated as follows: Where P represents the probability of subjects participating in regular physical exercise, D = 1 represents subjects in the regular physical exercise group, and x refers to 12 covariates.
The second analysis is used to estimate the average treatment effect on the treated (ATT) by the following model, which reflects the effect of regular physical exercise on the emotional, physical, and cognitive health of older adults.The significance of ATT is tested by a paired t-test (85).
Where Y 1 and Y 0 are the dependent variables of the matched samples in the regular physical exercise and non-physical exercise groups, respectively, P is the propensity value, D = 0 indicates the subjects without PE, and D = 1 indicates subjects with the regular physical exercise.As Y 0 |D = 1 cannot be directly observed, the establishment of the above model must satisfy the "unconfoundedness assumption, " which requires control factors associated with the participation of older adults in regular physical exercise.Finally, a boundary sensitivity analysis was conducted to explore to what extent the hidden selection bias can change the results of the treatment effectiveness obtained (84).
Covariate screening was conducted using the correlation analysis.To compare differences between the regular physical exercise group and the non-physical exercise group, the independent-sample t-test and the chi-squared test were employed.The significance level for all tests was set at 0.05.PSMATCH2 and NNMATCH in Stata 14.0 were performed to conduct the statistical analysis.

Covariate screening
Covariates in LPSM should include factors that are related to both independent and dependent variables in order to obtain a net estimate of the effect of independent variables on dependent variables (83,86).To screen out the covariates that are related to regular physical exercise, physical health, emotional health, and cognitive health (2018), the correlation coefficients were calculated, respectively.The results are presented in Supplementary Table S1.The level of regular physical exercise was found to be significantly related to 10 covariates, except for marital status (p = 0.16) and chronic disease (p = 0.07).However, these two covariates are significantly related to physical (p < 0.01), emotional (p < 0.01), and cognitive health (p < 0.01).Therefore, to investigate the association between regular physical exercise of older adults and the three health factors using the LPSM method, all of the aforementioned covariates should be included.

Descriptive statistics
Descriptive analyses were conducted in this study to examine two different types of samples and compare differences between two groups of samples.The study included 761 older adults who engaged in regular physical exercise and 1,031 older adults who did not engage in physical exercise.Supplementary Table S2 shows that there were significant differences in 12 covariates, except for marital status, between the two groups.Compared to the non-physical exercise group, the regular physical exercise group had a higher education status, a higher income level, more number of children, better interpersonal relationships, and other factors.These significant differences should be taken into account when evaluating the relationships between regular physical exercise and the emotional, physical, and cognitive health of older adults.

Propensity score matching analysis
A logistic regression model was conducted to analyze the relationship between regular physical exercise and 12 covariates.The results of the regression showed that the model had a strong overall explanatory power.Among the covariates, age (p = 0.02), household registration (p < 0.01), income level (p < 0.01), interpersonal relationships (p < 0.01), IADLs in the 2010 year (p = 0.03), depression in the 2010 year (p < 0.01), and cognitive ability in the 2010 year (p < 0.01) were found to be significant predictors of regular physical exercise among older adults (refer to Table 1).The regression results would be used to build a prediction model to calculate the propensity of an older adult participating in the regular physical exercise.The higher the propensity, the more likely the older adults do the regular physical exercise.

Matching and balanced tests
The calculated propensity was used to match 761 older adults with regular physical exercise and 1,031 older adults without PE.The nearest-neighbor matching method (1:1) was used, and finally, a total of 428 pairs of samples were successfully matched.As shown in Supplementary Table S3, the independent-sample t-test of each covariate between the control group and the treatment group was no longer significant (p > 0.05) after matching, which met the relevant recommended indicators.
In addition, to specifically demonstrate how the covariate balance between the two groups was improved by propensity score matching, the absolute values of the standardized differences before matching and after matching are displayed in Figure 2. The standard deviations (Figure 2, triangles) of all covariates after matching were improved substantially, and the absolute values of the standard deviation of all covariates were below 0.1, indicating that the assumption of the covariate balance is satisfied.

Average treatment effect on the treated
In this study, the bootstrap method was used to estimate the average treatment effect on the treated (ATT) and empirical standard error after propensity score matching.The estimated results show that older adults with regular physical exercise (treatment group) showed significantly lower depression levels (p < 0.01), higher physical health (p < 0.01), and higher cognitive health (p < 0.01) than older adults without PE (control group) after the bias of confounding factors was eliminated by LPSM.The depression level was significantly reduced by 0.187, and physical health and cognitive health were significantly improved by 0.542 and 1.032, respectively (see Table 2).In general, compared with the non-PE group, the physical health and cognitive health of older adults in the regular physical exercise group can be improved by 8.963 and 5.029%, respectively, and the depression level was reduced by 10.47%.

Sensitivity analysis
To further explore to what extent the hidden selection bias can change the results of the treatment effectiveness obtained, a sensitivity analysis with Rosenbaum bounds was used to assess the robustness of the estimates.Rosenbaum's method of sensitivity analysis relies on the sensitivity parameter gamma, which was defined as log odds of differential assignment due to unobserved confounders (84).The higher the gamma, the lower the sensitivity.Generally, if the gamma is greater than or equal to 2 and the p-value is significant, the results are insensitive to hidden bias (87).
As shown in Table 3, when gamma = 2, the treatment effects of regular physical exercise on physical health still is significant (p < 0.01).This finding indicated that this result was absolutely insensitive to hidden bias; when gamma = 1.8, the treatment effects of regular physical exercise on depression would no longer be significant (p = 0.08), which indicated that this result was possibly insensitive to hidden bias; and when gamma = 1.4, the treatment effects of regular physical exercise on cognitive health would no longer be significant (p = 0.97), which indicated that the result of the positive impact of regular physical exercise on cognitive health is not reliable.

Discussion
This longitudinal study is the first attempt to investigate the relationship between regular physical exercise among older adults and their emotional, physical, and cognitive health.The study applies the LPSM approach and utilizes data from two waves of the CFPS in 2010 and 2018.The present study aimed to address the significant public health concern of healthy aging, which has garnered increased attention globally.
Several specific findings emerged.The first major finding is that regular physical exercise among older adults was significantly associated with their physical health.Specifically, the scores of IADLs for older adults in the regular physical exercise group were 8.963% higher than those in the non-regular physical exercise group.This finding is in alignment with the findings from previous large-scale epidemiological studies, randomized controlled trials, and meta-analytic reviews (88-91), which demonstrated that regular physical exercise had a significant, beneficial association with the physical ability of older adults to carry out IADLs.Changes in the absolute value of the standard deviation of the covariates before and after matching 12 covariates.(92)(93)(94), another key and consistent finding from this study was that, compared to older adults without physical exercise, older adults with regular physical exercise could be expected to have a lower level of depression.This finding is supported by the theory of social interaction (95).Based on this theory, social interactions that may facilitate relationships tends to bolster social relationships and thereby can be beneficial for reducing the level of depressive symptoms among older adults.As regular physical exercise in daily life could be a great relationship builder for older adults, it can also help promote their interpersonal communication skill and opportunity; consequently, these harmonious interpersonal relationships undoubtedly help old adults to relieve their depressive symptoms.On the other hand, long-term regular physical exercise may affect physiological health, which is conducive to improve depressive symptoms, such as an increasing level of monoamine transmitters, thereby restoring normal function of the hypothalamus-pituitary-adrenal axis (96,97).Therefore, there is a need to encourage older adults with prior depressive illness, or at high risk of developing depressive illness, to participate in daily life physical exercise.
Third, although a statistically significant association was observed between regular physical exercise treatment and the cognitive health of older adults, the sensitivity analysis revealed that these results were sensitive to unobservable variables.This finding indicates that the potential positive association between regular physical exercise in later life and cognitive health may not be sufficient or valid.This result could explain the inconsistency across previous studies (98)(99)(100).For example, a review on physical exercise benefiting cognitive health during late periods of the lifespan found that 26 studies reported a positive correlation between physical exercise and maintenance or enhancement of cognitive health, five studies reported a dose-response relationship between physical activity and cognition, and one study showed a non-significant correlation (101).However, the majority of the evidence included in this review was of medium quality, and the overall risk of bias in the studies used in this review is moderately high (101).In future research, it is crucial to identify unobserved variables as potential confounding factors in order to improve the reliability of research conclusions.
Several limitations in this study need to be acknowledged in order to propose new directions for future investigations.First, the data used in this study are derived from the CFPS, which are second-hand survey data.These types of data have inherent limitations in terms of variable selection.Therefore, it is recommended that future research should develop specific research tools for assessing regular physical exercise, as well as physical, emotional, and cognitive health.Second, the covariates employed in this study may not encompass all the important potential confounding factors that could influence the relationship between regular physical exercise of older adults and their physical, emotional, and cognitive health.Factors such as drug use, life satisfaction, and intelligence level of older adults were not included in the analysis.Therefore, it is suggested that future research should investigate these factors as they could provide valuable insights.Third, although this study found a significant effect between regular physical exercise of older adults and their physical, emotional, and cognitive health, the sensitivity analysis revealed that the positive impact of regular physical exercise on cognitive health may not be entirely reliable.Consequently, future research should delve deeper into exploring the underlying mechanisms that link regular physical exercise with cognitive health.Fourth, it is important to note that regular physical exercise among older adults is a complex concept.This study only differentiated between older adults with regular physical exercise and sedentary older adults.Therefore, it did not investigate the specific types, frequencies, or intensities of regular physical exercise that may be most beneficial for self-care function, depression, and cognitive health of older adults.Future research should address this knowledge gap and provide more detailed insights into the optimal characteristics of regular physical exercise for older adults' overall wellbeing.
The major empirical findings discussed above have important implications.First, policymakers should focus on creating effective and appropriate policies that promote enjoyable and lifelong physical exercise among older adults.These policies can provide formal and informal guidelines for communities to plan, implement, and evaluate physical exercise programs for older adults.Second, communities should offer a variety of age-appropriate sports and recreation programs that are appealing to older adults.It is also important to ensure that the spaces and facilities meet or exceed the recommended safety standards.Third, older adults need to recognize that they are more likely to engage in sedentary behavior, which can have negative health outcomes.Therefore, it is crucial for older adults to promote and participate in suitable physical activities to reduce sedentary time.Additionally, it is worth noting that physical activity levels tend to decline with age, and older adults often struggle to meet recommended guidelines.However, small interventions that encourage replacing sedentary activities with active ones can be effective.For example, incorporating more active travel, such as walking to nearby shops for small purchases, can be a feasible approach.

Conclusion
In conclusion, our study suggests that regular physical exercise may have a significant impact on reducing depressive symptoms and improving the physical function of older adults.However, we did not find sufficient evidence to suggest that it affects cognitive function, even when considering various confounding factors.These findings emphasize the importance of promoting regular physical exercise among older adults as part of active aging strategies, aiming to lessen the burden on public health.
Depression, one of the most prevalent emotional problems among older adults, serves as an indicator of emotional health.In the 2010 and 2018 CFPS, depression was evaluated by the simplified version of the Centre for Epidemiologic Studies Depression Scale (CESD) (73): Six questions measure depressive mood: (a) I felt depressed; (b) I felt it was very hard to do anything; (c) My sleep was restless; (d) I felt lonely; (e) I felt sad;

FIGURE 1
FIGURE 1 Construction of panel data.Source: CFPS (65).China Family Panel Studies (CFPS) is a nationally representative, biannual longitudinal survey of Chinese communities, families, and individuals launched in 2010 by the Institute of Social Science Survey (ISSS) of Peking University, China.CFPS (65) means 2 waves of the China Family Panel Studies data in 2010 and 2018.

TABLE 1
Logistic regression estimates of regular physical exercise among older adults (N = 1792).
"a" represents the data of the variable from 2010 and the others from the data from 2018.

TABLE 2
ATT of regular physical exercise on health of older adults.Expanding and consolidating past research, this study was able to provide an accurate and compelling evidence for the positive role of regular physical exercise on physical function measures of IADLs by effectively eliminating sample-selection bias and mixed bias with the method of LPSM.Second, in line with the previous studies

TABLE 3
The results of the sensitivity analysis.