The Effect of Social Isolation on Sarcopenia: A Longitudinal Study among the Middle-Aged and Older Population in China

Sarcopenia is a common skeletal muscle disorder characterized by the progressive loss of muscle mass and strength that typically occurs from approximately 40 years of age and accelerates after 70 years [1]. According to a recent meta-analysis, the prevalence of sarcopenia was 11% in men and 9% in women aged over 60 years, with a higher prevalence in Asia than in other regions [2]. Sarcopenia is associated with adverse outcomes, such as disability, fall, fracture, lower health-related quality of life, hospitalization, and mortality [3, 4]. With increasing life expectancy, the prevalence of sarcopenia will continue to increase, resulting in a greater disease burden in the coming decades.

Identifying the risk factors is important to enhance healthy ageing by preventing the development of sarcopenia. The recognized risk factors for sarcopenia include age, nutritional deficiency, sedentariness, and chronic inflammation [5‒7]. However, little is known about the effects of social factors, such as social isolation, on sarcopenia. Social isolation refers to an objective state marked by the absence of social relationships, such as living alone and lack of contact with families and others; it is increasingly becoming a public health concern for middle-aged and older people [8]. We propose that social isolation affects the risk of sarcopenia through three mechanisms. Firstly, encouragement, support, or role models of healthy behaviours from family and friends can motivate middle-aged and older adults to adopt healthy lifestyles, such as adequate nutrient intake and exercise, to slow down the skeletal muscle loss [9, 10]. Secondly, having more family obligations and social activities can help improve mobility and motor function [11]. Thirdly, elevated psychological stress in socially isolated individuals alters the neuroendocrine system [12]. Social isolation may contribute to muscle loss by inducing chronic low-grade inflammation [6]. There is evidence that social isolation increases the risk of premature death, certain somatic diseases, and mental diseases [13‒15]. However, research investigating the association between social isolation and sarcopenia, particularly change in social isolation, is limited.

It is difficult to diagnose sarcopenia in a community setting. The gold standard is based on comprehensive measurements of skeletal muscle mass, muscle strength, and physical performance, which are time-consuming and require trained personnel. In addition, measurement of muscle mass using dual-energy X-ray absorptiometry is expensive and not widely available. Hence, the Asian Working Group for Sarcopenia (AWGS) [16] has introduced a simplified assessment of “possible sarcopenia” for the early identification of people with or at risk for sarcopenia. Possible sarcopenia is defined as either low muscle strength or low physical performance, with good diagnostic accuracy for sarcopenia [17]. Although several studies have shown that social isolation is associated with handgrip strength or physical performance [18, 19], no study has used possible sarcopenia as the outcome.

This study aimed to examine the effect of social isolation on possible sarcopenia based on the China Health and Retirement Longitudinal Study (CHARLS). We hypothesized that social isolation is associated with the onset of possible sarcopenia over a 4-year follow-up period. We also explored the effect of individual social isolation components (i.e., living alone, marital status, lack of contact with children and friends, and lack of social activities) and the change in social isolation on possible sarcopenia.

Detailed descriptions of CHARLS have been reported previously [20]. In brief, CHARLS was a study on ageing in China, which collected a nationally representative sample of residents aged 45 years or above and their spouses. Fieldwork started in 2011 with sample individuals from 28 provinces, 150 counties/districts, and 450 communities. A total of 17,708 participants were recruited at baseline (2011) and were followed up every two or 3 years until 2018.

In this study, we used data from the 2011 baseline and 2015 follow-up surveys. The follow-up survey after 2015 was not considered because it did not include the physical examination required for sarcopenia assessment. We excluded participants aged <45 years (n = 777), those with possible sarcopenia (n = 4,461) at baseline, and those with no information on possible sarcopenia (n = 4,725) at baseline. Those with missing values for social isolation (n = 5) or covariate factors (n = 410) at baseline were also excluded. Among the 7,730 participants free of possible sarcopenia at baseline, we further excluded those who were lost to follow-up (n = 846) and those with missing information on possible sarcopenia (n = 1,195) in the 2015 follow-up survey. The final analysis included 5,289 participants from 419 communities (Fig. 1).

Possible sarcopenia was defined as either low muscle strength (<18 kg in females, <28 kg in males) or low physical performance (five-time chair-stand test ≥12 s) [16]. Muscle strength was measured through handgrip strength. The participants were requested to squeeze the hand dynamometer (WL-1000, Nantong Yuejian Physical Measurement Instrument Co., Ltd., Nantong, China) as hard as possible for a few seconds. It was performed in a standing position with the elbows at a right angle. Each participant was tested twice for both hands, and the average maximum grip strength of each hand was used for diagnosis. Physical performance was measured once using the five-time chair-stand test. The participants were requested to stand up and sit down five times on a 0.47-m high chair, repeatedly and as quickly as possible, with their arms folded across their chest.

We constructed an index to measure the level of social isolation based on five items. Participants were assigned one point for each of the following five items: if they (1) lived alone; (2) were not married/cohabiting; (3) had no adult child or had less than monthly contact with their adult children; (4) had less than monthly interaction with friends; and (5) did not take part in any social activities over the past month (such as playing chess, Mahjong, or cards, performing volunteer or charity work, or participating in a community organization). The items correspond to those used to assess social isolation in a previous longitudinal study [21] and other studies based on CHARLS [15, 22]. Similar indices have been widely adopted in other studies on social isolation [23, 24]. The index ranges from 0 to 5, with high scores indicating a higher level of isolation. Participants were categorized according to social isolation as low (score 0), middle (score 1), or high (score ≥2) [25].

Based on previous studies [18, 26], we included demographics and health indicators as covariates. The demographics included age, sex, residence, and education level. Residence type was classified as urban or rural since there were urban-rural differences in social isolation and sarcopenia prevalence [27, 28]. Education level was divided into two categories based on the completion of elementary education as 40.1% of the participants had less than elementary education. Health indicators included the body mass index and chronic diseases. Chronic diseases were self-reported; they include hypertension, chronic lung diseases, heart diseases, psychiatric diseases, and arthritis, which are associated with a higher risk of possible sarcopenia [5]. Loneliness was also included since it is a subjective state related to but distinct from social isolation and adversely affects physical functioning [29]. Loneliness was assessed using one question: “how often did you feel lonely during the last week?” Loneliness was treated as a binary variable, i.e., not lonely (reported rarely or none of the time) and lonely (reported some, occasionally, or most of the time).

Baseline characteristics were described as counts and percentages. The χ² test was used to examine the differences in characteristics between the social isolation groups. Mixed-effects regression with a random intercept was employed to calculate the odds ratio (OR) and 95% confidence interval (CI) to examine the association between social isolation and possible sarcopenia. A binomial distribution with a logit link function was used to fit the mixed-effects model. The participants were considered as the fixed effect. The household nested within the community were treated as the random effect in models as participants in the same household or community might have similar characteristics. Three models were constructed in this study. The first model was a crude model; the second model included covariates including demographics and health indicators; based on the second model, loneliness was added to the third model. Stratified analyses were performed by sex and age. The association between each component of social isolation and possible sarcopenia was examined using the above models. In the first model, each component was placed separately into the model without any adjustment. Based on the first model, the second model was adjusted for all covariates. The third model was a fully adjusted model that simultaneously included five components.

In the secondary analyses, among 5,120 participants who had information on social isolation in follow-up, we examined social isolation change from baseline to follow-up and created three categories: stable (low-low, middle-middle, high-high), progressive (low-middle, low-high, middle-high), and regressive (middle-low, high-middle, high-low). We used a mixed-effects model to assess the association between change in social isolation and possible sarcopenia, adjusting for baseline social isolation and other covariates. Stratified analyses were performed according to baseline social isolation, sex, and age.

We conducted four types of sensitivity analyses. Firstly, we examined whether similar associations would be observed if social isolation was treated as a continuous variable in the above models. Secondly, we used multiple imputations by chained equations for covariates with missing values to create five imputation datasets for the reconstruction of the above models. Thirdly, to examine the potential ascertainment bias, we performed (i) logistic regressions to compare the characteristics of participants in the final dataset with those who were lost to follow-up and those who were followed up but had missing information on possible sarcopenia; (ii) multinomial logistic regression with the outcome categorized as non-sarcopenia (reference group), possible sarcopenia, lost to follow-up, and missing information on sarcopenia at follow-up. Lastly, to address potential reverse causation, we defined participants with borderline sarcopenia as those in the worst decile on grip strength (sex-specific) or five-time chair-stand test among the participants without possible sarcopenia. (i) We examined whether the association between social isolation and possible sarcopenia remained after removing participants with borderline sarcopenia. (ii) We explored the association between borderline sarcopenia at baseline and change in social isolation.

All analyses were conducted using R 4.1.0. The two-tailed p < 0.05 was considered the level of significance.

A total of 5,289 participants, aged 45–90 years, were included in the final analysis. Among them, 34.4% (1,819/5,289) were aged 60 years or above and 50.8% (2,685/5,289) were male. Table 1 summarizes the main characteristics of the participants according to social isolation. Overall, 12.8% (676/5,289), 30.9% (1,634/5,289), and 56.3% (2,979/5,289) of the participants reported low, middle, and high levels of social isolation, respectively.

After 4 years, possible sarcopenia was detected in 21.7% (1,146/5,289) of the participants. Table 2 shows the association between baseline social isolation and follow-up possible sarcopenia. The incidence of possible sarcopenia in the low, middle, and high social isolation groups was 13.2% (89/676), 21.1% (346/1,634), and 23.9% (711/2,979), respectively. In the univariate model, compared with the low social isolation group, the middle (OR = 1.77, 95% CI = 1.34–2.34, p < 0.001) and high social isolation groups (OR = 2.13, 95% CI = 1.63–2.79, p < 0.001) were associated with a higher risk of possible sarcopenia. The strength of such associations reduced after adjusting for demographics and health indicators, with a further slight reduction when loneliness was added to the model (middle social isolation group: OR = 1.53, 95% CI = 1.16–2.04, p = 0.003; high social isolation group: OR = 1.65, 95% CI = 1.26–2.18, p < 0.001; p for trend = 0.001). In the fully adjusted model (Table 2), loneliness was associated with an increased risk of possible sarcopenia (OR = 1.23, 95% CI = 1.03–1.46, p = 0.021). The results of the full regression model are presented in online supplementary Table S1 at www.karger.com/doi/10.1159/000529443. No multicollinearity was found in the two adjusted models (all variance inflation factor <2). When stratified by sex or age, increased social isolation was associated with an increased risk of possible sarcopenia in each sex or age stratum (all p for trend <0.05). No interaction was found between social isolation and age, sex (online Suppl. Table S2), or other characteristics (online Suppl. Table S3).

Table 3 shows the associations of the five social isolation components with possible sarcopenia. The crude model showed that living alone, being not married/cohabiting, lack of contact with children, and lack of social activities were associated with possible sarcopenia. After adjusting for other factors, the adjusted OR for being not married/cohabiting, lack of contact with children, and lack of social activities remained significant. The fully adjusted model also showed that not married/cohabiting (OR = 1.58, 95% CI = 1.19–2.10, p = 0.002), lack of contact with children (OR = 1.86, 95% CI = 1.21–2.85, p = 0.004), and lack of social activities (OR = 1.26, 95% CI = 1.04–1.53, p = 0.019) were significantly associated with possible sarcopenia.

Table 4 shows the association between change in social isolation and possible sarcopenia. Compared with participants who maintained social isolation levels from baseline to follow-up, those with progressive social isolation were associated with a greater risk of possible sarcopenia (OR = 1.51, 95% CI = 1.17–1.95, p = 0.001). Stratified analyses by baseline social isolation showed that progression from low or middle to high social isolation elevated the risk of possible sarcopenia; however, the reduction in social isolation was not associated with a reduced risk of possible sarcopenia. No interaction was found between the change in social isolation and age or sex. However, when stratified by age, there was no association between progressive social isolation and the risk of possible sarcopenia among participants aged 60 years or older (online Suppl. Table S4).

We performed four types of sensitivity analyses. The first analysis treated social isolation as a continuous variable, and the results were similar to those of the categorical analyses (online Suppl. Table S5). Secondly, the results from five imputation dataset analyses did not show changes in interpretation (online Suppl. Table S6). Thirdly, the results showed that participants who were male, had a higher level of loneliness, had chronic diseases, and lived in urban areas were more likely to be lost to follow-up (online Suppl. Table S7), and male, younger, and urban participants were more likely to have missing sarcopenia information without loss of follow-up (online Suppl. Table S8). The results of the multinomial logistic regression analysis were consistent with those of the main analysis (online Suppl. Table S9). Lastly, the association between social isolation and possible sarcopenia remained significant after removing participants with borderline sarcopenia (online Suppl. Table S10), and no association between borderline sarcopenia at baseline and change in social isolation was found (online Suppl. Table S11).

This longitudinal study explored the effect of social isolation on possible sarcopenia. Social isolation was found to be associated with the development of possible sarcopenia after 4 years of follow-up. These associations remained significant after adjusting for baseline demographics, health indicators, and loneliness.

In this study, we used both muscle strength (grip strength) and physical performance (five-time chair-stand test) to diagnose possible sarcopenia. Compared to the low social isolation group, the middle and high social isolation groups were associated with a higher risk of possible sarcopenia. Several studies using muscle strength or physical performance rather than possible sarcopenia as outcomes produced results consistent with our study, that is, social isolation was associated with lower grip strength or poorer physical performance. Yu et al. [18] found that baseline high social isolation predicted grip strength decline over 4 years in Chinese men. However, a study in Japan showed no association between social isolation and a decline in handgrip strength. This might be due to the short follow-up period (1 year) and small sample size (166 subjects). Cruz et al. [30] used the Short Physical Performance Battery to assess physical performance by combining gait speed, balance, and muscle function. The results revealed that social isolation was associated with a decline in physical functioning after 9 years. Other studies have yielded similar results [19, 26]. The possible mechanisms, as stated in the introduction section, could be decreased physical activity and increased sedentary time [31], which are recognized risk factors of sarcopenia; increased risk of inadequate diet among socially isolated people [32], which might reduce the ability to synthesize muscle proteins [33]; and increased proinflammation as social isolation is associated with increased C-reactive protein [34], which is related to reduced body function and muscle loss [35, 36]. However, the exact mechanism requires further investigation.

Results on individual components of social isolation showed that being not married/cohabiting, lack of contact with children, and lack of social activities were associated with an increased risk of possible sarcopenia. These findings could inform interventions to reduce social isolation and prevent sarcopenia in these groups. Adult children should strengthen their connections with their ageing parents through face-to-face, phone, or video calls to provide emotional support. Community activities for middle-aged and older adults should be organized so that they participate more in social activities. Finally, health education on sarcopenia prevention should be provided for middle-aged and older populations, especially single, divorced, and widowed people who lack family and social engagement. No association was observed between lack of contact with friends and possible sarcopenia, which is possibly because in China, middle-aged and older adults have a strong bond with family [37]. Age-friendly cities (AFC), a concept developed by the World Health Organization (WHO) [38], is a significant initiative at the social level in response to social isolation. The WHO Global Network of AFC has included 1,439 cities worldwide [39]. Social participation, an important domain of AFC, emphasizes a series of accessible and affordable activities for older people and intergenerational integration.

Our study showed that compared with participants who maintained a stable social isolation level, there was a higher risk of possible sarcopenia in those with progressive social isolation. However, there was no decrease in the risk of possible sarcopenia in participants with regressive social isolation. This suggests that the effect of social isolation on possible sarcopenia might be accumulative and irreversible; this indicates the importance of minimizing social isolation and preventing its progression, for the prevention of possible sarcopenia. We noted that the association between progressive social isolation and possible sarcopenia was not significant among people aged 60 years or older, which might be due to the small sample size in the subgroup because no interaction between change in social isolation and age was found. Future studies with a bigger sample size are required to confirm this association in older adults.

This study also found that baseline loneliness also increased the risk of possible sarcopenia after 4 years. A study with 12 years of follow-up observed a longitudinal association between baseline loneliness and decreased physical functioning [29], which is consistent with that observed in our study. However, in our study, the measurement of loneliness was based only on a single question regarding the participants’ perception in the last week. Therefore, more evidence is required to explore the association between loneliness and sarcopenia.

To the best of our knowledge, this is the first study to investigate the effect of social isolation and change in social isolation on the development of possible sarcopenia. The longitudinal design helped address the reverse causality between social isolation and the development of sarcopenia. Another advantage of this study is that CHARLS is a nationally representative study of middle-aged and older Chinese residents. However, our study has the following limitations. Firstly, as the appendicular skeletal muscles of participants were not measured in CHARLS, only possible sarcopenia could be used as the outcome of this study. However, possible sarcopenia has good diagnostic accuracy for sarcopenia, with a sensitivity of 0.893 and 0.921, and a specificity of 0.990 and 0.870 for men and women, respectively [17]. In addition, muscle strength is a better predictor of adverse outcomes than muscle mass [40]. Secondly, since the physical examination manual did not specify the time interval between the handgrip strength tests, it was unclear whether the participants’ muscles had sufficient time to recover before the second test. Thus, we considered the average of the maximum grip strength of each hand following the AWGS recommendation [16]. Thirdly, the proportion of participants lost to follow-up (including missing information on possible sarcopenia at follow-up) in our study was relatively high. There were differences in age, sex, loneliness, and residence between the included participants and those lost to follow-up, which may have led to ascertainment bias. However, the association between social isolation and possible sarcopenia remained significant in multinomial logistic regression that took into account participants lost to follow-up. Lastly, although a 4-year follow-up might be a short duration for developing sarcopenia, the association between social isolation and possible sarcopenia was observed in this study, which provides a springboard for future research. Future studies with longer follow-up periods are required to confirm this finding.

Social isolation is associated with an increased risk of possible sarcopenia, and progressive social isolation further elevates the risk. The most vulnerable groups are middle-aged and older people who live alone, are not socially active, and lack contact with their children. This highlights the importance of social contacts in sarcopenia development and provides new directions for the prevention of sarcopenia.

We thank the China Health and Retirement Longitudinal Study team for providing data and all respondents for their contribution.

The CHARLS study obtained ethics approval from the Institutional Review Board at Peking University (IRB00001052-11015). All participants signed the informed consent. The analysis was approved by the Survey and Behavioural Research Ethics Committee of the Chinese University of Hong Kong (SRBE-21-0483).

The authors have no conflicts of interest to declare.

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Study design and conception: P.K.M.P. and B.H.K.Y.; data analysis, interpretation, and manuscript drafting: P.H.; and results discussion and critical revision: S.Y.S.W., D.Z., J.W., and R.Y. All authors read and approved the final manuscript.

The data that support the findings of this study are openly available in CHARLS at http://charls.pku.edu.cn/. Further enquiries can be directed to the corresponding author.