The Parallel Mediation Effects of Depression, Well-Being, and Social Activity on Physical Performance and Frailty in Community-Dwelling Middle-Aged and Older People

Background Frailty refers to a decline in an elderly person's physical, psychological, and social functioning, making them sensitive to stressors. Because frailty is caused by a variety of factors, including certain demographic characteristics, understanding the mediating factors that affect frailty in the elderly is critical. Purpose To provide evidence about the relationship between depression, well-being, social activity, physical performance, and frailty among older adults. Materials and Methods The study used secondary data from Taiwan's Long-term Study of Aging (n = 7,622), excluding people with severe dementia. The chi-square test and Spearmen's coefficient correlation were used to assess the relationship between the demographic variables and frailty. Nonparametric bootstrapping analysis was used to test whether depression, well-being, and social activity are parallel mediators of the relationship between physical performance and frailty. This study was approved by Fu Jen Catholic University (FJU-IRB No. C110040). Results The overall frailty prevalence was 13.9%. We calculated a mean score and standard deviation for each measurement in this study. The correlation found low-to-moderate positive and negative statistically significant correlations between the variables. A significant, moderately negative relationship was found between physical performance and frailty that correlated with three potential mediating factors. The path indicated that lower physical performance scores and higher depression scores are more likely to be associated with frailty. Conclusion Older adults who are depressed are more likely to become frail. Adults who are more socially active and report greater well-being are less likely to become frail. Therefore, further research should design and test a comprehensive intervention for older adults in community settings that addresses all three factors, aimed at increasing well-being and social activity while also treating depression.


Introduction
Meeting the particular needs of frail older adults in community settings would improve health outcomes and quality of life while also lowering health care costs in any country [1,2]. Frailty refers to a multidimensional loss of human functioning that may be physical, psychological, or social [3]. Moreover, frailty increases as people age, from a prevalence of 30% among people aged 60 and older to 40% among those over 70, so it presents a burden for individuals and families in the community [4]. Te presence of frailty, poor health, and multiple comorbidities results in a high risk of adverse outcomes, such as falls and disability [5]. Terefore, identifying frailty is important in providing clinical care for older adult community-dwelling populations.
Previous studies have shown that demographic factors contribute to frailty, such as being female and increasing age [6,7], while medical factors also contribute to frailty, such as cognitive impairment, sarcopenia, falls, and institutionalization [8][9][10][11][12]. However, mediation studies have revealed that other factors associated with frailty also include physical condition, psychological status, and social activity.
Te literature documents associations between frailty and these three additional variables. For example, a study that examined the relationship between frailty and physical performance (i.e., movement, behavior, and body composition) found that it was mediated by a person's physical activity, with sedentary time and moderate-to-vigorous physical activity acting independently as mediators [13]. A study of the association between frailty and cognitive function found that it was mediated by psychological distress; those sufering higher levels of psychological distress had higher rates of frailty [14]. Another study found that the association between frailty and loneliness was mediated by social activity and engagement [15]. However, a cross-sectional study of the relationship between frailty and social activity among older adults in Japan during the COVID 19 pandemic found no link between frailty and being hindered in social activity [16]. Tis fnding might be explained by the unique conditions of the pandemic, which limited social activity for frail and not-so-frail people alike [16].
Te presence of comorbidities is an understudied predictor of frailty, so the evidence is limited. Studies of the relationship between frailty and physical disease in community-dwelling older adults have mostly used simple and serial mediation; few studies have used parallel mediation, which can include more than two variables [17]. However, parallel mediation analysis has yielded interesting results in identifying symptoms related to frailty [17]. Te principles of parallel mediation used in this study are described in the work of Kane and Ashbaugh in 2017 and Hayes and Preacher in 2014 [18,19]. We evaluated the magnitude of the degree correlation according to the work of Hopkins [20], namely, robust (r � 0.7-0.8), strong (r � 0.5-0.7), moderate (r � 0.3-0.4), small (0.1-0.2), and trivial (r < 0.1) Tis study used parallel mediation analysis to evaluate the efects of depression, well-being, and social activity on the relationship between frailty and physical performance.

Study Design and Sample.
Tis study used secondary data from a cross-sectional study, the 2015 Taiwan Longitudinal Study on Aging (TLSA). Te participants were thus not directly involved in this study. Te total TLSA sample data consisted of 8,300 older adults in Taiwan who were divided into two age groups, those aged 50 to 64 and those aged 65 to 85. After excluding older adults with severe dementia, the total number of participants in the study was 7,622. Participants' data were anonymized to protect privacy and human rights according to the Declaration of Helsinki guidelines on research with human subjects. Tis study was approved by Fu Jen Catholic University (FJU-IRB No. C110040).

Te Instrumental Activities of Daily Living (IADL)
Scale. Te IADL scale, created by Lawton and Brody, has been used to measure physical performance since 1969 and consists of nine items representing independent living skills.
Te Chinese version of the IADL has a Cronbach's alpha range of 0.82 to 0.92 [21]. Te scale measures difculty performing nine tasks, each with two possible responses, yes (able) or no (not able). "Yes" answers are given one point, and "no" answers are assigned a value of zero; the total scores thus range from zero to nine. Te TLSA and this study used the IADL to measure physical performance as an independent variable. . Te GDS-15 short form is used to measure depression; it was introduced by Yesavage in 1982. It includes ffteen items with two possible responses: yes or no. Depression levels are categorized as follows: scores from 0 to 4 indicate normal; scores from 5 to 8 indicate mild depression; scores from 9 to 11 indicate moderate depression; and scores from 12 to 15 indicate severe depression. For 10 of the items, positive responses are assigned 1 point, and for 5 items, negative responses are assigned 1 point (items 1, 5, 7, 11, and 13). Te Cronbach's alpha coefcient for the total scale is 0.80 [22].

World Health Organization-5 (WHO-5) Scale.
Te WHO-5 scale is used to evaluate well-being in older adults. It consists of 5 items; each item can have a score from 0 to 5, with the overall raw score ranging from 0 to 25. All scores are then multiplied by 4, with the highest total score being 100. Cronbach's alpha coefcient for this scale ranged from 0.81 to 0.86 [23].

Te Social Activity Scale (SAS).
Te SAS was developed in Japan to measure social activity among older adults in community settings [24]. Items asked about activities at home or in the community, with responses "yes" and "no" assigned point values of one and zero, respectively. Te highest possible score for social activity is 15, which indicates having leisure time and social life in the community. Tis 15item scale has Cronbach's alpha of 0.791 [24].

2.2.5.
Frailty. Tis study measured frailty by evaluating the physical, psychological, and social conditions of older adults in community settings using the Tilburg Frailty Indicator [3]. It has a total possible score of 15, with scores of 5 or more indicating frailty [3]. For this study, the variables representing frailty were constructed from the TLSA 2015 data [25,26].
Control variables were demographic characteristics such as age, gender, education, marital status, and economic condition, as well as diagnoses in the past year and the number of chronic diseases.

Statistical Analysis.
Analysis was performed using the process function of SPSS version 22.0 statistical software. In this study, 9%-12% of participants had missing data on demographic characteristics and were excluded from this study. Te continuous variables are displayed using means and standard deviations (SDs), and the categorical variables are displayed using case numbers and percentages (%). Te chi-square test was used to assess the relationship between frailty and the demographic variables, which were then assessed for signifcance using the crosstab function of SPSS. Mentioning the p values in the demographic characteristics is that p value is important for developing a rigorous statistical analysis to consider whether any variables in the demographic data could potentially afect the study's results. Data were examined to assess the normality of the distribution using the Shapiro-Wilk test, a histogram, and Q-Q plots. Since the data were not normally distributed, nonparametric, and dichotomous, Spearman's correlation coefcient was used to investigate the statistical relationship between the variables [27]. Spearman's correlation analysis was conducted to determine the relationships among fve variables, physical performance, depression, well-being, physical activity, and frailty, all of which were statistically signifcant with p values less than 0.05.
Nonparametric bootstrapping analysis [19] was used to test depression, well-being, and social activity as potential parallel mediators of the relationship between frailty and physical performance. Te mediation is signifcant if the CI is 95% and the bias-corrected lower and upper limits for indirect efects do not include zero [19]. Tree mediation models were developed to test the hypothesis that depression, social activity, and well-being afect the relationship between physical performance and frailty. Parallel mediation efects were tested using model four of the bootstrap method of Preacher and Hayes. We used the PROCESS function version 4.0 in SPSS version 22 to analyze mediation.

Demographic Characteristics.
Descriptive demographic characteristics of the participants (n = 7,622) are shown in Table 1. Te prevalence of frailty was 13.9%, and 86.1% of older adults were robust. A total of 87.78% of the participants had no chronic diseases; 82.71% of the participants had a high school education or below; and 78.75% of the participants were satisfed with their economic status. Moreover, the results showed statistically signifcant correlations between frailty and gender, age, education, marital status, having a disease diagnosed in the past year and chronic diseases. Te exception was the association between frailty and satisfaction with economic status, which was not statistically signifcant (p = 0.196).

Correlations between Frailty and Physical Performance,
Depression, Well-Being, and Social Activity. Spearman's correlation shows the relationships among physical performance, depression, well-being, social activity, and frailty ( Table 2). We calculated mean scores and standard deviations for each of these variables. Spearman's correlation found both positive and negative, statistically signifcant (p < 0.01), low-to-moderate relationships with frailty. Physical performance had a signifcant negative association with depression (rho = −0.245) and frailty (rho = −0.562), but it was positively associated with well-being (rho = 0.216) and social activity (rho = −0.335; p < .01).

Testing the Partial Mediation Models.
Partial mediation analysis found that both direct and indirect mediating efects were signifcant (p < 0.001), with 95% confdence intervals in bootstrap analysis (Table 3). Model 1 showed that depression negatively mediated the relationship between physical performance and frailty. Models 2 and 3 tested positive for mediation. Te efect of depression on frailty was positive (β = 0.20; p < 0.001), but well-being (β = −0.19; p < 0.001) and social activity (β = −0.18; p < 0.001) had no efect on frailty.

Testing Parallel Mediation Models.
In the path, the direct efect of (a) physical performance on depression was negative and statistically signifcant (β = −0.34; p < 0.001), (b) the direct efect of depression on frailty was positive and signifcant (β = 0.20; p < 0.001), while (c) the direct efect of physical performance on frailty is negative and statistically signifcant (β = −1.03; p < 0.001) (Figure 1). Te path indicates that older adults who have lower scores on physical performance and higher scores on depression are more likely to be frail.

Demographic Characteristics.
Tis article investigated the potential mediating factors in the relationship between physical performance and frailty. Te results indicated that the prevalence of frailty was 13.9%. More than threequarters of the participants had no chronic disease, and a similar proportion had lower education levels. However, both chronic disease and education were signifcantly correlated with frailty in this study. Tis fnding is consistent with previous studies in Taiwan that also found that level of education and chronic disease were signifcantly correlated with frailty [26]. Other evidence indicates that a lack of education is linked to poor cognitive performance, functional disability, and weakness, all characteristics of frailty [28][29][30]. Another study found that patients who become frail due to chronic diseases did not difer from the other groups in terms of functional, cognitive, or psychological status. Older adults with chronic diseases had a higher comorbidity burden and more frailty [31]. Terefore, in order to get clear data about frailty, patients with dementia were excluded from this study.

Association between Study Variables.
Tis study demonstrated associations between frailty and physical performance, depression, well-being, and social activity ( Table 2). Te association between physical performance and depression was negative and signifcant, as was the relation between physical performance and frailty. Tese fndings are consistent with previous studies, which found that physical performance was negatively associated with depression scores [13,29,32]. Other studies have found four functional Current Gerontology and Geriatrics Research limitations associated with frailty: (1) physical functioning, (2) activities of daily living, (3) instrumental activities of daily living (IADL), and (4) constraints on social participation [33]. Weak upper-and lower-body physical performance was associated with depression among older adult women [34]. Te results of this and other studies indicate that physical impairments are crucial predictors of depression and frailty in Taiwan. In turn, depression might be a mediating factor in the relationship between physical performance and frailty.
In contrast, the association between physical performance and well-being was signifcantly positive but low, while well-being and frailty had a low negative correlation. Previous studies found that physical activity has a positive efect on well-being among older adults [35]. For example, social media used as an IADL signifcantly predicts social well-being among older adults in the twenty-frst century [36]. Others have shown that aspects of well-being, such as a sense of autonomy, control, purpose, and fulfllment, as well as happiness and pleasure, protect against the development of frailty [37]. In addition, cultural diferences in how people defne well-being may be explained by diferences in norms about appropriate emotions and the importance of mental health [38]. Moreover, the correlation between physical performance and well-being may difer by region or country [39]. Furthermore, unique individual diferences may impact these relationships; for example, maladaptive psychological responses would negatively impact psychological well-being, which might then contribute to frailty [40]. In this study, well-being was found to be a mediating factor in the relationship between physical performance and frailty; therefore, we should pay attention to the mental well-being of older adults in order to reduce the risk of frailty.
In this study, social activity was found to be a mediating factor in the relationship between physical performance and frailty. Our results showed a low but signifcant positive correlation between social activity and physical performance; in other words, participants who were more socially active had a better physical performance. Previous studies  have likewise found a signifcant positive correlation between physical performance and social activity [41,42]. One systematic review indicated that the greater the level of social activity, the greater the level of physical activity [42]. Tat review study also demonstrated a positive association between physical activity and social support among older adults who received social support from family members and friends. Higher levels of physical and social activity were, in turn, related to a lower incidence of depression and frailty among the older adults [42]. Tis connection between more social activity and lower depression may be explained by two phenomena: social activities are known to improve cognitive function and loneliness causes depression in older adults [43][44][45][46]. Meanwhile, in this study, social activity and frailty were moderately negatively correlated, meaning that those who had more social activity were less likely to be frail. Finally, this study found a moderately negative correlation between physical performance and frailty that was afected by three mediating factors, namely, depression, well-being, and social activity. Next, we tested these mediating factors in the relationship between physical performance and frailty for parallel mediation.

Parallel Mediation of the Relationship between Physical
Performance and Frailty. On the relationship between physical performance and frailty, three mediation models were tested for each mediator (depression, well-being, and social) ( Table 3), while parallel mediation (Figure 1) summarized the fndings. Te mediators of depression, well-being, and social activity acted as independent mediators in the relationship between physical performance and frailty in this study. Terefore, these fndings might be used to supplement previous research on frailty and older adults in Taiwan. Our results confrmed that older adults with higher levels of depression tended to be more  Current Gerontology and Geriatrics Research depressed than older people who were less depressed. Our study demonstrated that the efect of physical performance on frailty was partially mediated by depression among older adults. Tis fnding is in line with previous studies that showed that older adults with more depression had a higher probability of sufering from frailty [14]. Moreover, depression is not only associated with frailty [13] but also it is afected by weak upper-and lower-body physical performance [34] and sedentary time [28,32]. Tis may explain how depression mediates the efect of physical performance on frailty. In this study, older adults with impaired physical performance and a high level of depression had more severe frailty. Te results of this study also confrmed that older adults with higher levels of well-being were signifcantly less frail than older adults with lower levels of well-being. Previous studies have likewise found that well-being as a psychological state, which includes a sense of autonomy, happiness, and pleasure, is protective against the development of frailty [37]. However, physical performance is also relevant to positive mental well-being [35]. Moreover, cultural diferences, such as diferences in norms about emotions considered characteristic of mental health, may afect the measurement of well-being in older people [39]. Our fndings indicate that wellbeing mediates the efect of physical performance on frailty. Not surprisingly, older adults with good physical performance and high levels of well-being were more robust in this study. In contrast, maladaptive psychological responses negatively impact psychological wellbeing and might contribute to frailty [39]. Terefore, health professionals should be aware that maintaining a strong sense of psychological well-being among older adults is important for their physical health.
Regarding the social domain in the older adult, living alone, social relation, social support, and social participation are related to frailty [47][48][49]. Moreover, a group of frailty experts agreed to include social activity in the development of an integral conceptual model of frailty [47]. Because it is important for an integrated human view, the social domain cannot be ignored in frailty assessment [48]. Te social domains are related to (1) physical functioning, (2) ADL, (3) IADL, and (4) constraints on participation [50]. One study noted that the efects of social participation on cognitive function were greater in younger males than those in older males in Taiwan [43]. In sum, these studies have provided no evidence that social activity is a mediator of the relationship between physical performance and frailty. Terefore, the evidence related to the efect of social activity on physical performance and frailty is still at an early stage of understanding in this study.

Conclusion
To the best of our knowledge, the mediating factors in the relationship between physical performance and frailty have been discussed in several articles up to the present date. Te limitations of the mediators of well-being are still unclear. Tis study provided preliminary cross-sectional support for a mediational model in which physical performance and frailty potentially explain the relationships among physical, psychological, and social factors among older adults in the community. To prevent frailty among older adult participants, comprehensive interventions should be designed for future research.

Data Availability
Te data that support the fndings of this study are available from Health Data Science Center, Taiwan, but restrictions apply to the availability of these data, which were used under license for the current study, and so are not publicly available. Te data are, however, available from the corresponding author upon reasonable request and with permission of the Taiwan Ministry of Health and Welfare.

Conflicts of Interest
Te authors declare that they have no conficts of interest.