Skip to main content
NCBI home page
The Journals of Gerontology Series B: Psychological Sciences and Social Sciences logoLink to The Journals of Gerontology Series B: Psychological Sciences and Social Sciences
. 2023 Jan 26;78(5):809–818. doi: 10.1093/geronb/gbad013

The Longitudinal Association Between Frailty, Cognition, and Quality of Life in Older Europeans

Wei Hu 1, Jiadong Chu 2, Yixian Zhu 3, Xuanli Chen 4, Na Sun 5, Qiang Han 6, Tongxing Li 7, Zhaolong Feng 8, Qida He 9, Jun Wu 10, Yueping Shen 11,✉
Editor: Rodlescia Sneed
PMCID: PMC10413812  PMID: 36702742

Abstract

Objectives

Evidence on the association between frailty and quality of life (QoL) is mostly limited to cross-sectional studies. Thus, the temporal order and potential mechanisms of this association are largely unknown. Our study examines both the directionality of this association and the role of cognition in this association in longitudinal data.

Methods

Cross-lagged panel models were employed to examine the temporal relationship between frailty and QoL, as well as cognition’s role among 19,649 older adults in Europe. Frailty, QoL, and cognition were assessed using the health deficit index, CASP-12, and 3 standard cognitive tests, respectively.

Results

We observed a bidirectional association between frailty and QoL and their dynamics. High initial levels of frailty predicted poorer QoL later and vice versa (β = −0.151 and −0.052, p < .001). The early change in frailty predicted the late change in QoL, and vice versa (β = −0.093 and −0.061, p < .001). Frailty or its early change drives this interrelationship. Cognition at Wave 5 partially mediated frailty’s effect at Wave 4 on QoL at Wave 6 (indirect effect: β = −0.005, 95% confidence interval = −0.006, −0.004).

Discussion

Our findings supported that early prevention of frailty and its risk factors may have more influential protective effects on later physical and mental health, as well as the need for ongoing screening for mental health in aging population. Also, the maintenance of good cognitive performance may help interrupt this possible vicious cycle linking frailty and QoL decline.

Keywords: Bidirectional association, Cognition, Frailty, Mediation, Quality of life

Background

In clinical terms, frailty is a reduced physiological reserve associated with aging (Hoogendijk et al., 2019). It is characterized by reduced resilience to stressors, increased vulnerability, and reduced ability to maintain internal homeostasis due to the dysregulation of multiple physiological systems (Bergman et al., 2007; Fried et al., 2001). The prevalence of frailty continues to increase with age. A recent systematic review suggested that more than a quarter of older adults may be affected by frailty (Veronese et al., 2021). Frailty has been proven to be a key predictor of a range of adverse health outcomes, such as falls, disability, frequent hospitalizations, depression, and mortality (Hoogendijk et al., 2019). All these may impair the quality of life (QoL) of frail older people.

QoL is an individual’s perceptions of their life status concerning their goals, expectations, standards, and concerns within the context of the culture and value systems (1993). It is a multidimensional concept including life satisfaction, psychological well-being, and positive feelings and functioning (Mol et al., 2009). With the increase in population aging worldwide, the life expectancy of older adults has increased significantly (Kojima et al., 2016). However, longevity does not represent a high QoL in later life (Kojima et al., 2016). Therefore, maintaining a good QoL is usually considered a crucial part of successful aging (Rowe & Kahn, 1998). QoL is increasingly recognized as an essential indicator of the health burden of frailty on the older adult population (Kojima et al., 2016). Frailty is a dynamic and potentially reversible risk factor among various physical, psychological, and social factors that worsen the QoL of older adults (Hoogendijk et al., 2019). Consequently, understanding the reciprocal relationship between frailty and QoL will help develop targeted interventions to improve overall health.

Previous studies have consistently shown an inverse association between frailty and QoL; however, some questions remained unanswered. First, most previous studies utilized small samples and cross-sectional designs, limiting their ability to test causality and the representativeness of findings (Gobbens et al., 2013; Masel et al., 2009). Second, existing studies using longitudinal design emphasized the one-way association between frailty and QoL, specifically, the association of baseline frailty with subsequent QoL (Veronese et al., 2022). For example, the results of a 10-year follow-up study suggested that frailty at baseline is a predictor of subsequent lower QoL (Veronese et al., 2022). Other longitudinal studies supported the view that QoL acts as a protective factor in the progression of frailty (Gale et al., 2014). For instance, Gale et al. found that maintaining a good QoL in later life can lower the risk of frailty’s onset (Gale et al., 2014). Based on previous findings, a potential bidirectional association between frailty and QoL may exist. Furthermore, understanding the interrelationship between changes in frailty and QoL from a dynamic perspective is critical. If this relationship exists, an early indication of frailty could provide a timely warning of the subsequent decline in QoL and vice versa. Recognizing this association will enable clinicians to intervene early when they detect a decline in any health dimension. Hence, conducting a large-scale longitudinal study using cross-lagged panel data would be valuable for disentangling the association’s direction and dynamics.

A growing amount of research has focused on the relationship between frailty and QoL in older adults; however, the psychosocial mechanisms underlying this longitudinal association remain unclear. Cognitive decline and frailty are common, closely related complications of aging. Moreover, the bidirectional relationships between frailty and cognition and between cognition and QoL also demonstrated that cognition might mediate the bidirectional relationship between frailty and QoL (Aranda et al., 2011; Auyeung et al., 2011; Doba et al., 2012; Hussenoeder et al., 2020; Phyo et al., 2021; Samper-Ternent et al., 2008). Several prospective epidemiological studies showed that higher levels of frailty predict cognitive decline in older adults and vice versa (Aranda et al., 2011; Auyeung et al., 2011; Doba et al., 2012; Samper-Ternent et al., 2008). Existing research also supported the latter bidirectional association (i.e., the potentially reciprocal link between cognition and QoL). For instance, a prospective study conducted in Germany reported mild cognitive impairment as a predictor of future poor QoL in older adults (Hussenoeder et al., 2020). A longitudinal study of 19,106 community-dwelling older adults found that higher QoL at baseline predicted a lower risk of subsequent cognitive decline and dementia (Phyo et al., 2021). It is theoretically possible that cognitive performance mediates the bidirectional frailty–QoL pathway in late life. However, this mediational hypothesis in older adults remains untested.

Therefore, this study explores the bidirectional association between frailty and QoL and further examines whether a reciprocal association exists between changes in frailty and QoL in older adults. Additionally, it evaluates whether cognition mediates the potential bidirectional relationship utilizing a three-wave cross-lagged panel design based on a large nationally representative sample from the Survey of Health, Aging, and Retirement in Europe (SHARE).

Method

Study Participants

The present longitudinal study is based on a sample of 19,649 older adults aged ≥50 years from 12 European countries who participated in Waves 4 (2011), 5 (2013), and 6 (2015) of the SHARE. A detailed description of SHARE is available in Börsch-Supan et al. (2013). We primarily focused on cross-lagged relationships over time. Therefore, the sample was restricted to participants who provided three repeated measurements for frailty, cognition, and QoL. The data gathered in Waves 4, 5, and 6 were selected for analysis because the study variables were measured three times across these waves. Figure 1 shows the screening procedure of the study participants.

Figure 1.

Figure 1.

Open in a new tab

Flow chart of the study participants. Notes: QoL = quality of life; SHARE = Survey of Health, Aging, and Retirement in Europe.

The Ethics Council of the Max Planck Society and the University of Mannheim provided ethical approval for the SHARE project. All participants signed an informed consent form.

Quality of Life

The CASP-12 scale, including four domains—control, autonomy, self-realization, and pleasure—was used to assess QoL in SHARE (Santini et al., 2020). Each domain contained three questions which were assessed using a 4-point Likert scale, with respondents’ responses including never (4 points), rarely (3 points), sometimes (2 points), and often (1 point). Therefore, the total score ranged from 12 to 48. Higher scores indicated better QoL. In addition, participants were considered to possess a good QoL when their CASP score was above the population median (Veronese et al., 2022). Cronbach’s alpha for the CASP-12 scale at Wave 4 was 0.81.

Frailty

The two most commonly used measures of frailty are the categorical frailty phenotype (FPP) model developed by Fried et al. and the continuous frailty index (FI) model based on deficit accumulation (Fried et al., 2001; Rockwood et al., 2005). The FI is designed to be more sensitive in identifying patients with moderate and severe frailty than the FPP (Kulminski et al., 2008). Moreover, the continuous nature of FI makes it more suitable for longitudinal follow-up studies that explore changes in health status over time (Cesari et al., 2014). Consequently, the FI was used to measure frailty in this study. Initially, 92 deficit parameters including symptoms (e.g., cognition and depression), signs (e.g., tremors), disease states, and disabilities were used to define frailty (Mitnitski et al., 2001). Subsequent studies have proven that these parameters can be reduced to a minimum of around 30 without losing predictive validity (Song et al., 2010). The 44- and 40-item FI containing cognitive and depression-related deficits, respectively, have been validated in SHARE as valid predictors of adverse health outcomes (Mayerl et al., 2020; Romero-Ortuno, 2013). For example, Mayerl et al. constructed a 44-item FI by excluding depression-related deficits to explore the relationship between frailty and depression (Mayerl et al., 2020). This study is based on this previous research, a 46-item FI was constructed by excluding two cognitive-related deficits and re-adding four depression-related deficits to examine the relationship between frailty, cognition, and QoL in all three waves (see Supplementary Table 1). Most (44/46) variables were dichotomous, with deficits for each item coded as “1” or no deficit as “0.” Ordered response variables were assigned an intermediate score between 0 and 1. The FI score (score range 0–1) was equal to the total number of deficits reported divided by all the deficits assessed (Mitnitski et al., 2001). In line with previous literature, 0.25 was the frailty threshold (FI ≥ 0.25; Song et al., 2010).

Cognitive Function

The three cognitive tests including immediate and delayed recall and executive function were employed to assess cognitive function in the current sample (Foverskov et al., 2018). The Ten-Word Recall test assessed respondents’ ability to recall 10 common words immediately and after a delay (range 0–20). Executive function was evaluated using a verbal fluency test. Participants were asked to name as many different animals as possible in 1 minute (range 0–67). The sum of these tests was used as the total cognitive scores for each wave, with higher scores indicating better cognitive function. We standardized the total cognitive scores for measurement consistency across waves (Bourassa et al., 2015).

Covariates

Sociodemographic characteristics (sex [male/female], age, education attainment), health-related behaviors (smoking [current/former/never], alcohol consumption [current/never], physical activity), and health conditions (body mass index [BMI], and the number of chronic noncommunicable diseases [NCDs]) were adjusted for in the current study. All are potential common causes of frailty and poor QoL (Gobbens & van Assen, 2017; Xu et al., 2021). Education level was assessed according to the International Standard Classification of Educational Degrees using the following three categories: low (0–2), medium (3–4), and high (5–6) (Foverskov et al., 2018). Physical activity was measured based on self-reported frequency of engaging in moderate or vigorous exercise, classified as never engaging in vigorous and moderate physical activity (0) or other physical activity (1). NCDs were determined based on the question, Have you ever been told by a doctor that you have any of the following 13 diseases, such as high blood pressure, diabetes, high blood cholesterol, or stroke? (Santini et al., 2020). Thus, the number of NCDs ranged from 0 to 13 and was divided into four categories: none (0), one (1), two (2), and three or more (3). The EURO-D scale in SHARE uses 12 binary items to measure depression (Mayerl et al., 2020). However, the entries measuring FI contained four depression-related categories (sad or depressed, hopelessness, fatigue, and lack of enjoyment). Therefore, we did not consider depression a covariate to reduce multicollinearity in the primary analysis.

Statistical Analysis

Descriptive analysis of baseline characteristics used frequency (%) for qualitative data and mean ± standard deviations (SDs) or medium (Quartile 1 and Quartile 3) for quantitative data. Wilcoxon (t) and Chi-square tests were employed to examine the differences in baseline characteristics according to the presence of frailty or QoL status at Wave 6. Spearman’s correlation test was utilized to analyze the relationship among frailty, cognition, and QoL at the three time points.

Cross-lagged structural equation models were created to test the bidirectional relationships among frailty and QoL at the selected waves (Figure 2), by constructing three models. Model 1 was adjusted for sociodemographic factors. Model 2 was further adjusted for health-related factors. Model 3 included all covariates in Model 2, plus health conditions. In the model system, the cross-lagged effect between frailty and QoL (a1 and a2, b1 and b2) was emphasized. Cross-lagged models estimate the unidirectional causal effect between the same variables and estimate the directional influence of one variable on another over time, and thus are considered to be appropriate for examining potential reciprocal relationships in longitudinal data analysis (Xu et al., 2022). Additionally, several sensitivity analyses were performed to validate the results’ robustness. First, we tested the temporal relationship between frailty and QoL by restricting the cross-lagged effects across waves to be equal. Considering the impact of mental health disorders (mainly depression) on frailty and QoL (Borges et al., 2021; Sivertsen et al., 2015), we adjusted the depression composed of the remaining eight items on the basis of Model 3 to explore the interrelationship. Then, we further excluded four depression-related deficits from the 46-item FI while adjusting for depression as a covariate to perform sensitivity analyses. Given the vulnerability of frailty and QoL to age, we conducted subgroup analyses by dividing age into two groups (50–70 and ≥70 years). Then, based on prior research, cross-lagged latent change models were constructed to examine the dynamics between early changes in frailty and QoL and late changes in QoL and frailty (Figure 3; Sha et al., 2022). In the model, we focused on the cross-lagged effect of early change in frailty or QoL on late change in another variable (e1, e2).

Figure 2.

Figure 2.

Open in a new tab

Standardized path diagram of cross-lagged model between frailty and QoL, SHARE (N = 19,649), 2011–2015. Notes: For the sake of brevity, all covariates and residuals were estimated in the analysis but not shown in the diagram. Model adjusted for age, sex, education level, smoking, drinking, physical activities, BMI, and number of chronic diseases. BMI = body mass index; FI4, FI5, and FI6 = frailty at Waves 4, 5, and 6; QoL4, QoL 5, and QoL 6 = quality of life at Waves 4, 5, and 6.

Figure 3.

Figure 3.

Open in a new tab

Standardized path diagram of cross-lagged model between change in frailty and QoL, SHARE (N = 19,649), 2011–2015. Notes: BMI = body mass index; △early-FI = FI at Wave 5 minus FI at Wave 4; △late-FI = FI at Wave 6 minus FI at Wave 5; △early-QoL = QoL at Wave 5 minus QoL at Wave 4; △late-QoL = QoL at Wave 6 minus QoL at Wave 5. Covariates including age, sex, education level, smoking, drinking, physical activities, BMI, and number of chronic diseases, and residuals were estimated in the analysis but not shown in the diagram. Only cross-lagged path coefficients are shown in the diagram. ***p < .001.

Next, a cross-lagged mediation model was built to estimate the indirect effect of cognition (f1 × g1 and f2 × g2) on the longitudinal relationship between frailty and QoL through the coefficient product method (Figure 4; Bentley et al., 2013). A bias-corrected bootstrap method (10,000 draws) assessed the indirect effect’s significance (Mackinnon et al., 2004). If the 95% bootstrap confidence interval (CI) did not include zero, the indirect effect was significant. Standardized path coefficients and 95% CI were reported to compare the magnitude of the predicted effects and determine which variable has a greater effect on the other. The Sobel test examined the significance of the differences in standardized path coefficients.

Figure 4.

Figure 4.

Open in a new tab

Standardized path diagram of cross-lagged mediation model, SHARE (N = 19,649), 2011–2015. Notes: Covariates including age, sex, education level, smoking, drinking, physical activities, BMI, and number of chronic diseases, and residuals were estimated in the analysis but not shown in the diagram. Only cross-lagged path coefficients are shown in the diagram. FI4, FI5, and FI6 = frailty at Waves 4, 5, and 6; Cog4, Cog5, and Cog6 = cognition at Waves 4, 5, and 6; QoL4, QoL 5, and QoL 6 =: quality of life at Waves 4, 5, and 6. ***p < .001. **p < .01. *p < .05.

A range of fit statistics, including standardized root-mean-square residual (SRMR), root-mean-square error of approximation (RMSEA), comparative fit index (CFI), goodness-of-fit index (GFI), and incremental fit index (IFI), assessed each model’s plausibility. When the CFI, GFI, and IFI values were greater than or equal to 0.90, the RMSEA and SRMR values were less than or equal to 0.08, and the fit of the model was acceptable (Bentler, 1990).

All analyses were performed in SAS version 9.4 (SAS Institute Inc., Cary, NC), excluding cross-lagged path analysis, which was implemented in AMOS (SPSS, IBM, Armonk, NY; Version 24.0). For all analyses, p < .05 was statistically significant.

Results

Baseline Characteristics and Correlations

Of the 19,649 participants included in the analysis, 55.7% were females. The mean age (SD) was 64.9 (8.8) years. Compared with nonfrail older adults, those who were frail at the follow-up (Wave 6) were older, female, less educated, less physically active, more likely to be current smokers and drinkers, had more NCDs, had a higher BMI, had a poorer QoL, and had a greater FI at baseline (Table 1). Moreover, similar findings were obtained when baseline characteristics were described in terms of QoL at Wave 6 (see Supplementary Table 2).

Table 1.

Baseline Characteristics of Study Participants According to Frailty Status at Wave 6

Characteristics Total (n = 19,649) Nonfrail (n = 17,637) Frail (n = 2,012) p
Age, years (mean ± SD) 64.9 ± 8.8 64.3 ± 8.5 70.3 ± 9.7 <.001
Sex, n (%) <.001
 Male 8,694(44.3) 8,031(45.5) 663(33.0)
 Female 10,955(55.7) 9,606(54.5) 1,349(67.0)
Level of education, n (%) <.001
 Low 6,851(34.9) 5,802(32.9) 1,049(52.1)
 Medium 8,067(41.1) 7,363(41.7) 704(35.0)
 High 4,731(24.1) 4,472 (25.4) 259(12.9)
Smoking status, n (%) <.001
 Current smoker 3,581(18.2) 3,215(18.2) 366(18.2)
 Former smoker 3,146(16.0) 2,813(16.0) 333(16.6)
 Nonsmoker 12,922(65.8) 11,609(65.8) 1,313(65.3)
Drinking status, n (%) <.001
 Nondrinker 12,659(64.4) 11,788(66.8) 871(56.7)
 Current drinker 6,990(35.6) 5,849(33.2) 1,141(43.3)
BMI, kg/m2 (mean ± SD) 26.9 ± 4.6 26.6 ± 4.4 29.3 ± 5.7 <.001
Physical activities, n (%) <.001
 Never 1,120(5.7) 693(3.9) 427(21.2)
 Other 18,529(94.3) 16,944(96.1) 1,585(78.8)
Number of NCDs, n (%) <.001
 0 4,912(25.0) 4,833(27.4) 79(3.9)
 1 5,972(30.4) 5,674(32.2) 298(14.8)
 2 4,259(21.9) 3,866(21.9) 429(21.3)
 ≥3 4,470(22.7) 3,264(18.5) 1,206(60.0)
FI at Wave 4 (mean ± SD) 0.10 ± 0.10 0.26 ± 0.13 0.09 ± 0.08 <.001
QoL score, medium (Q1, Q3) at Wave 4 39(35, 43) 40(35, 43) 34(29, 38) <.001
Open in a new tab

Notes: BMI = body mass index; FI = frailty index; NCDs = chronic noncommunicable diseases; Q1 = Quartile 1; Q3 = Quartile 3; QoL = quality of life; SD = standard deviation.

Frailty was inversely related to cognition and QoL (ranging from −0.197 to −0.481) at each wave and across waves. A higher FI at Wave 4 was associated with lower levels of cognition function at Wave 4, and lower QoL at Wave 5. Cognitive function was significantly positively correlated with QoL at each time point (ranging from 0.198 to 0.650; all p < .001; see Supplementary Table 3).

Bidirectional Association Between Frailty and QoL

The standardized path coefficients for the final model (Model 3) are diagrammed in Figure 2. The model yielded a good data fit (GFI = 0.981; IFI = 0.967; CFI = 0.967; SRMR = 0.035; RMSEA = 0.067). The correlations between frailty and QoL at each time point were statistically significant and showed inverse relationships. The subsequent autoregressive effects of frailty and QoL reached significance (all p < .01). For the cross-lagged effects, frailty at Wave 4 was negatively and significantly associated with QoL at Wave 5 (β = −0.151, p < .001), and vice versa (β = −0.052, p < .001). Likewise, frailty at Wave 5 predicted poor QoL at Wave 6 (β = −0.217, p < .001) and vice versa (β = −0.081, p < .001). Furthermore, the path associations and model fitting were still stable when different covariates were adjusted (Supplementary Table 4). To examine the magnitude of the strength of the cross-lagged relationship between frailty and QoL, we further compared differences in standardized path coefficients, and found that the predictive effect of frailty on subsequent QoL was significantly stronger than that of QoL on subsequent frailty (all p < .001). The equivalence of constraint cross-lagged effects did not affect the bidirectional association between frailty and QoL (β = −0.178, p < .001 vs β = −0.083, p < .001; see Supplementary Figure 1). When adjusting for the depression composed of the remaining eight items as a covariable or excluding four depression-related deficits from FI, we found that the results of the sensitivity analysis were consistent with the preliminary analysis (Supplementary Tables 5 and 6). The results of the subgroup analysis by age group were consistent with the main findings (see Supplementary Figure 2).

The Bidirectional Association Between Changes in Frailty and QoL

Figure 3 shows the standardized estimate results of the cross-lagged association between the dynamic changes in frailty and QoL (Supplementary Table 7 displays full-model results). Baseline frailty was a significant negative predictor of early and late changes in QoL (β = −0.177 and −0.173, respectively, p < .001). The late change in QoL was significantly predicted by an early change in frailty (β = −0.093, p < .001). This result suggests that older adults with high and worsening frailty experienced a faster decline in QoL during follow-up. Baseline QoL inversely predicted early and late changes in frailty (β = −0.122 and −0.070, respectively, p < .001). The early change in QoL predicted the late change in frailty (β = −0.061, p < .001). Thus, frailty deteriorated more slowly among older adults with high baseline levels and improving QoL. To examine the magnitude of the cross-lagged relationship between changes in frailty and QoL, we further compared differences in standardized path coefficients (e1 and e2). The predictive effect of the early change in frailty on the late change in QoL was significantly stronger than it was in the opposite direction (p < .01).

The Mediating Role of Cognitive Function in the Longitudinal Relationship Between Frailty and QoL

Figure 4 depicts the standardized path coefficients for the cross-lagged mediation model (Supplementary Table 8 presents a full-model summary). The model fits the data well (GFI = 0.980; IFI = 0.972; CFI = 0.972; SRMR = 0.035; RMSEA = 0.056). The findings indicated that cognition at Wave 5 partially mediated the relationship between frailty at Wave 4 and QoL at Wave 6 (f1 × g1: β = −0.005, 95% CI: −0.006, −0.004; h1: β = −0.198, bootstrap 95% CI: −0.212, −0.185). Conversely, the indirect effect of QoL at Wave 4 on frailty at Wave 6 through cognition at Wave 5 was not significant (f2 × g2: β = −0.0002, bootstrap 95% CI: −0.0006, 0.0002). Overall, these results suggest that cognition played a partial mediating role in the effect of frailty on subsequent QoL, but not vice versa.

Discussion

Our study is the first known to demonstrate a negative bidirectional association between frailty and QoL based on a large nationally representative longitudinal cohort study of older adults aged 50 years and older in Europe. Moreover, older adults who experienced increasing frailty over the first 2 years of the study also experienced a faster decline in QoL in the last 2 years and vice versa. Frailty or its early change drives this interrelationship. Even after adjusting for various potential confounders and performing a series of subgroup and sensitivity analyses, the direction and magnitude of the bidirectional association remained robust. Furthermore, this study explored the mediating role of cognition in this association. The exacerbation of early frailty may lead to a reduced QoL in later life, partly because it impairs cognitive function, which was previously unknown. These findings provide new insights into the direction and mechanisms of the relationship between frailty and QoL. In the context of the global population’s transition to aging, improving the QoL of the frail elderly population, reversing their frailty, and enabling them to live healthy, happy lives is a primary health goal of all governments.

The dynamic, bidirectional association between frailty and QoL is supported by previous studies, although these associations have only been demonstrated separately and unidirectionally. Most of these studies supported that frailty predicts poor QoL (Gobbens & van Assen, 2014; Masel et al., 2009; Siriwardhana et al., 2019; Veronese et al., 2022). Frailty is associated with disturbances of the nervous, endocrine, and immune systems and skeletal muscle loss, thereby increasing the risk of various adverse outcomes, ultimately leading to worsening QoL (Gobbens & van Assen, 2014; Siriwardhana et al., 2019). However, improving QoL is also an effective strategy to prevent incident frailty (Gale et al., 2014; Ostir et al., 2004). The link between psychological well-being and plasma inflammatory markers, which are risk factors for the deterioration of frailty, provides strong support for the physiological mechanism by which poor QoL predicts later frailty (Friedman et al., 2007; Hubbard & Woodhouse, 2010). To our knowledge, only one prospective study among older adults in the United Kingdom found a possible bidirectional association of frailty with QoL (Gale et al., 2014). Nonetheless, this U.K. study highlighting QoL’s prediction of incident frailty in older adults did not simultaneously test this bidirectional association. Instead, it tested associations in only one direction at a time through different statistical methods and thus may be limited in clarifying the temporal nature of the relationship. The cross-lagged panel model utilizing repeated measures is suitable for examining the temporal relationship between variables (Xu et al., 2022). Notably, we observed that initial frailty or early changes had a greater impact on subsequent QoL or its late changes than in the opposite direction. A longitudinal study examining the lead–lag relationship between physical and mental health in older adults proposed that physical health has a greater impact on mental health, which supports our findings (Luo et al., 2020). Changes in physical health have a more rapid and pronounced effect on mental health. In contrast, the adverse impact of mental health on physical health gradually emerges throughout the life span (Sha et al., 2022). Therefore, early prevention of frailty may have a more significant protective effect on overall older adult health.

Additionally, the current study identified a mediating mechanism for this association. Frail older adults can maintain subsequent good QoL through improved cognitive function. Earlier prospective studies have indicated that higher frailty at baseline is associated with cognitive decline at follow-up, robustly supporting the first half of the longitudinal mediated relationship found in our study (Auyeung et al., 2011; Mitnitski et al., 2011; Samper-Ternent et al., 2008). Several biological mechanisms could explain this link. For instance, testosterone, an important factor in preventing the development of frailty, is thought to have a protective effect on cognition by promoting synaptic plasticity in the hippocampus and regulating amyloid beta accumulation (Robertson et al., 2013). Also, insulin resistance is associated with an increased risk of developing frailty and with poorer cognitive function (Robertson et al., 2013). A systematic review reported that cognitive decline was strongly associated with lower QoL, which aligns with the second half of our mediating effect (Hill et al., 2017). Cognitive decline is associated with adverse outcomes such as self-reported poor health, inadequate social networks, anxiety, and depression, thereby affecting an individual’s QoL (Mol et al., 2009; Pusswald et al., 2015).

Strengths and Limitations

Several strengths of this study should be mentioned. First, based on a large nationally representative sample of older adults in Europe, this study was the first to elucidate the temporal relationship between frailty and QoL and their dynamics in the context of longitudinal data. Second, frailty was first found to be a longitudinal precursor in the bidirectional association. Early screening and prevention of frailty may have greater benefits for the physical and mental health of older adults. Third, we found that cognition partially mediates the adverse influence of frailty on QoL using a longitudinal mediation design. Longitudinal data with repeated measurement might provide more information about the temporal order of independent, mediator, and outcome variables, crucial prerequisites for clarifying causality involved in mediation analysis (Bentley et al., 2013).

Some limitations in this study should also be noted. First, frailty, cognition, and QoL were based on self-reported data. Although the measurement methods for these variables have been previously validated (Han et al., 2021; Mitnitski et al., 2001; Santini et al., 2020), measurement errors and deviations from common methods may still occur. Moreover, we examined only one possible mechanism and found that a partially mediated effect of cognition, although statistically significant, could only explain a small proportion of the total effect of frailty on QoL. This result suggests that other substantial unexplored mediators along the causal chain from frailty to QoL exist. Further studies are warranted to test these mediators in the future. Third, the nature of observational studies makes it necessary to infer causality cautiously. However, the longitudinal design with cross-lagged allows us to test the plausibility of the proposed causality model with greater certainty than with cross-sectional studies. Fourth, the study only used the FI method based on deficit accumulation to measure frailty. Thus, the findings may vary due to the definition and measurement of frailty. In the future, using frailty as defined by other measurement methods (such as the Fried phenotype approach or the Edmonton Frail Scale) will be necessary to confirm the robustness of the research results (Fried et al., 2001; Rolfson et al., 2006). Fifth, the cross-lagged model involving all individuals measured at three time points resulted in a significant reduction in sample size, which means that selection bias may have affected the study results. Sixth, the study did not control for any mental health disorders as covariates, which may have affected the findings. However, in the sensitivity analyses, we further controlled for depression as a covariate and found that the study results remained robust. Finally, we must consider the impact of missing data and loss of follow-up on representativeness. Nevertheless, the comparison of the analytic sample with the total population at baseline is not significantly different (see Supplementary Table 9).

Conclusion

Building on a 5-year cross-lagged longitudinal data, our study detected a possible negative bidirectional relationship between frailty and QoL and their dynamics among older adults in Europe. Furthermore, we observed that frailty played a dominant role in this bidirectional relationship. The prospective frailty-to-QoL relationship is partially mediated by cognition function. Accordingly, early monitoring and prevention of frailty and its risk factors may have more influential protective effects on the overall health of older adults, as well as the need for ongoing screening for mental health conditions in aging population. Also, the maintenance of good cognitive performance may help break this possible vicious cycle linking frailty and QoL decline.

Supplementary Material

gbad013_suppl_Supplementary_Tables
Click here for additional data file. (258.5KB, docx)

Acknowledgments

We thank the German Ministry of Education and Research, the Max Planck Institute for the Advancement of Science, and the U.S. National Institute on Aging (U01_AG09740-13S2, P01_AG005842, P01_AG08291, P30_AG12815, R21_AG025169, Y1-AG-4553-01, IAG_ BSR06-11, OGHA_04-064, HHSN271201300071C) for additional funding for the development of the SHARE project. We would like to thank Editage (www.editage.cn) for English language editing.

Contributor Information

Wei Hu, Department of Epidemiology and Biostatistics, School of Public Health, Medical College of Soochow University, Suzhou, China.

Jiadong Chu, Department of Epidemiology and Biostatistics, School of Public Health, Medical College of Soochow University, Suzhou, China.

Yixian Zhu, School of Radiation Medicine and Protection, Soochow University, Medical College of Soochow University, Suzhou, China.

Xuanli Chen, Department of Epidemiology and Biostatistics, School of Public Health, Medical College of Soochow University, Suzhou, China.

Na Sun, Department of Epidemiology and Biostatistics, School of Public Health, Medical College of Soochow University, Suzhou, China.

Qiang Han, Department of Epidemiology and Biostatistics, School of Public Health, Medical College of Soochow University, Suzhou, China.

Tongxing Li, Department of Epidemiology and Biostatistics, School of Public Health, Medical College of Soochow University, Suzhou, China.

Zhaolong Feng, Department of Epidemiology and Biostatistics, School of Public Health, Medical College of Soochow University, Suzhou, China.

Qida He, Department of Epidemiology and Biostatistics, School of Public Health, Medical College of Soochow University, Suzhou, China.

Jun Wu, Department of Oncology, The Third Affiliated Hospital of Soochow University, Changzhou, Jiangsu, China.

Yueping Shen, Department of Epidemiology and Biostatistics, School of Public Health, Medical College of Soochow University, Suzhou, China.

Funding

This work was supported by the National Natural Science Foundation of China (project number 81973143).

Conflict of Interest

All authors reported no potential conflicts of interest.

Author Contributions

W. Hu, J. Wu, and Y. Shen: study design. W. Hu, J. Chu, N. Sun, X. Chen, Y. Zhu, T. Li, Z. Feng, Q. He, and Y. Shen: data collection and management. W. Hu and Q. Han: data analyses. All authors: critical revision and final approval of the manuscript.

Data Availability

Because the data are publicly available (https://share-eric.eu/data/) in the Harmonized SHARE data set and Codebook, Version E as of October 2019 developed by the Gateway to Global Aging Data, we do not share data.

Ethical Approval and Consent to Participate

The SHARE project received ethical approval from the Ethics Committee of the Max Planck Society for the Advancement of Science, and all participants provided informed consent at recruitment. All methods were performed in accordance with the relevant guidelines and regulations.

References

  1. Aranda, M. P., Ray, L. A., Snih, S. A., Ottenbacher, K. J., & Markides, K. S. (2011). The protective effect of neighborhood composition on increasing frailty among older Mexican Americans: A Barrio advantage? Journal of Aging and Health, 23, 1189–1217. doi: 10.1177/0898264311421961 [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Auyeung, T. W., Lee, J. S., Kwok, T., & Woo, J. (2011). Physical frailty predicts future cognitive decline—A four-year prospective study in 2737 cognitively normal older adults. The Journal of Nutrition, Health and Aging, 15, 690–694. doi: 10.1007/s12603-011-0110-9 [DOI] [PubMed] [Google Scholar]
  3. Bentler, P. M. (1990). Comparative fit indexes in structural models. Psychological Bulletin, 107, 238–246. doi: 10.1037/0033-2909.107.2.238 [DOI] [PubMed] [Google Scholar]
  4. Bentley, J. P., Brown, C. J., McGwin Sawyer, G. P., Allman, R. M., & Roth, D. L. (2013). Functional status, life-space mobility, and quality of life: A longitudinal mediation analysis. Quality of Life Research, 22, 1621–1632. doi: 10.1007/s11136-012-0315-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Bergman, H., Ferrucci, L., Guralnik, J., Hogan, D. B., Hummel, S., Karunananthan, S., & Wolfson, C. (2007). Frailty: An emerging research and clinical paradigm—Issues and controversies. The Journals of Gerontology, Series A: Biological Sciences and Medical Sciences, 62, 731–737. doi: 10.1093/gerona/62.7.731 [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Borges, M. K., Aprahamian, I., Romanini, C. V., Oliveira, F. M., Mingardi, S. V. B., Lima, N. A., Cecato, J. F., Petrella, M., & Oude Voshaar, R. C. (2021). Depression as a determinant of frailty in late life. Aging and Mental Health, 25, 2279–2285. 
doi: 10.1080/13607863.2020.1857689 [DOI] [PubMed] [Google Scholar]
  7. Börsch-Supan, A., Brandt, M., Hunkler, C., Kneip, T., Korbmacher, J., Malter, F., Schaan, B., Stuck, S., & Zuber, S. (2013). Data resource profile: The Survey of Health, Ageing and Retirement in Europe (SHARE). International Journal of Epidemiology, 42, 992–1001. doi: 10.1093/ije/dyt088 [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Bourassa, K. J., Memel, M., Woolverton, C., & Sbarra, D. A. (2015). A dyadic approach to health, cognition, and quality of life in aging adults. Psychology and Aging, 30, 449–461. doi: 10.1037/pag0000025 [DOI] [PubMed] [Google Scholar]
  9. Cesari, M., Gambassi, G., van Kan, G. A., & Vellas, B. (2014). The frailty phenotype and the frailty index: Different instruments for different purposes. Age and Ageing, 43, 10–12. doi: 10.1093/ageing/aft160 [DOI] [PubMed] [Google Scholar]
  10. Doba, N., Tokuda, Y., Goldstein, N. E., Kushiro, T., & Hinohara, S. (2012). A pilot trial to predict frailty syndrome: The Japanese Health Research Volunteer Study. Experimental Gerontology, 47, 638–643. doi: 10.1016/j.exger.2012.05.016 [DOI] [PubMed] [Google Scholar]
  11. Foverskov, E., Glymour, M. M., Mortensen, E. L., Holm, A., Lange, T., & Lund, R. (2018). Education and cognitive aging: Accounting for selection and confounding in linkage of data from the Danish Registry and Survey of Health, Ageing and Retirement in Europe. American Journal of Epidemiology, 187, 2423–2430. doi: 10.1093/aje/kwy162 [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Fried, L. P., Tangen, C. M., Walston, J., Newman, A. B., Hirsch, C., Gottdiener, J., Seeman, T., Tracy, R., Kop, W. J., Burke, G., & McBurnie, M. A. (2001). Frailty in older adults: Evidence for a phenotype. The Journals of Gerontology, Series A: Biological Sciences and Medical Sciences, 56, M146–M1 56. doi: 10.1093/gerona/56.3.m146 [DOI] [PubMed] [Google Scholar]
  13. Friedman, E. M., Hayney, M., Love, G. D., Singer, B. H., & Ryff, C. D. (2007). Plasma interleukin-6 and soluble IL-6 receptors are associated with psychological well-being in aging women. Health Psychology, 26, 305–313. doi: 10.1037/0278-6133.26.3.305 [DOI] [PubMed] [Google Scholar]
  14. Gale, C. R., Cooper, C., Deary, I. J., & Aihie Sayer, A. (2014). Psychological well-being and incident frailty in men and women: The English Longitudinal Study of Ageing. Psychological Medicine, 44, 697–706. doi: 10.1017/s0033291713001384 [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Gobbens, R. J., Luijkx, K. G., & van Assen, M. A. (2013). Explaining quality of life of older people in the Netherlands using a multidimensional assessment of frailty. Quality of Life Research, 22, 2051–2061. doi: 10.1007/s11136-012-0341-1 [DOI] [PubMed] [Google Scholar]
  16. Gobbens, R. J., & van Assen, M. A. (2014). The prediction of quality of life by physical, psychological and social components of frailty in community-dwelling older people. Quality of Life Research, 23, 2289–2300. doi: 10.1007/s11136-014-0672-1 [DOI] [PubMed] [Google Scholar]
  17. Gobbens, R. J. J., & van Assen, M. (2017). Associations between multidimensional frailty and quality of life among Dutch older people. Archives of Gerontology and Geriatrics, 73, 69–76. doi: 10.1016/j.archger.2017.07.007 [DOI] [PubMed] [Google Scholar]
  18. Han, F. F., Wang, H. X., Wu, J. J., Yao, W., Hao, C. F., & Pei, J. J. (2021). Depressive symptoms and cognitive impairment: A 10-year follow-up study from the Survey of Health, Ageing and Retirement in Europe. European Psychiatry, 64, e55. doi: 10.1192/j.eurpsy.2021.2230 [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Hill, N. L., McDermott, C., Mogle, J., Munoz, E., DePasquale, N., Wion, R., & Whitaker, E. (2017). Subjective cognitive impairment and quality of life: A systematic review. International Psychogeriatrics, 29, 1965–1977. doi: 10.1017/s1041610217001636 [DOI] [PubMed] [Google Scholar]
  20. Hoogendijk, E. O., Afilalo, J., Ensrud, K. E., Kowal, P., Onder, G., & Fried, L. P. (2019). Frailty: Implications for clinical practice and public health. Lancet, 394, 1365–1375. doi: 10.1016/s0140-6736(19)31786-6 [DOI] [PubMed] [Google Scholar]
  21. Hubbard, R. E., & Woodhouse, K. W. (2010). Frailty, inflammation and the elderly. Biogerontology, 11, 635–641. doi: 10.1007/s10522-010-9292-5 [DOI] [PubMed] [Google Scholar]
  22. Hussenoeder, F. S., Conrad, I., Roehr, S., Fuchs, A., Pentzek, M., Bickel, H., Moesch, E., Weyerer, S., Werle, J., Wiese, B., Mamone, S., Brettschneider, C., Heser, K., Kleineidam, L., Kaduszkiewicz, H., Eisele, M., Maier, W., Wagner, M., Scherer, M., Önig, H. H., … Riedel-Heller, S. G. (2020). Mild cognitive impairment and quality of life in the oldest old: A closer look. Quality of Life Research, 29, 1675–1683. doi: 10.1007/s11136-020-02425-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Kojima, G., Iliffe, S., Jivraj, S., & Walters, K. (2016). Association between frailty and quality of life among community-dwelling older people: A systematic review and meta-analysis. Journal of Epidemiology and Community Health, 70, 716–721. doi: 10.1136/jech-2015-206717 [DOI] [PubMed] [Google Scholar]
  24. Kulminski, A. M., Ukraintseva, S. V., Kulminskaya, I. V., Arbeev, K. G., Land, K., & Yashin, A. I. (2008). Cumulative deficits better characterize susceptibility to death in elderly people than phenotypic frailty: Lessons from the Cardiovascular Health Study. Journal of the American Geriatrics Society, 56, 898–903. doi: 10.1111/j.1532-5415.2008.01656.x [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Luo, M. S., Chui, E. W. T., & Li, L. W. (2020). The longitudinal associations between physical health and mental health among older adults. Aging and Mental Health, 24, 1990–1998. doi: 10.1080/13607863.2019.1655706 [DOI] [PubMed] [Google Scholar]
  26. Mackinnon, D. P., Lockwood, C. M., & Williams, J. (2004). Confidence limits for the indirect effect: Distribution of the product and resampling methods. Multivariate Behavioral Research, 39, 99. doi: 10.1207/s15327906mbr3901_4 [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Masel, M. C., Graham, J. E., Reistetter, T. A., Markides, K. S., & Ottenbacher, K. J. (2009). Frailty and health related quality of life in older Mexican Americans. Health and Quality of Life Outcomes, 7, 70. doi: 10.1186/1477-7525-7-70 [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Mayerl, H., Stolz, E., & Freidl, W. (2020). Frailty and depression: Reciprocal influences or common causes? Social Science and Medicine, 263, 113273. doi: 10.1016/j.socscimed.2020.113273 [DOI] [PubMed] [Google Scholar]
  29. Mitnitski, A., Fallah, N., Rockwood, M. R., & Rockwood, K. (2011). Transitions in cognitive status in relation to frailty in older adults: A comparison of three frailty measures. The Journal of Nutrition, Health and Aging, 15, 863–867. doi: 10.1007/s12603-011-0066-9 [DOI] [PubMed] [Google Scholar]
  30. Mitnitski, A. B., Mogilner, A. J., & Rockwood, K. (2001). Accumulation of deficits as a proxy measure of aging. Scientific World Journal, 1, 323–336. doi: 10.1100/tsw.2001.58 [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Mol, M. E., van Boxtel, M. P., Willems, D., Verhey, F. R., & Jolles, J. (2009). Subjective forgetfulness is associated with lower quality of life in middle-aged and young-old individuals: A 9-year follow-up in older participants from the Maastricht Aging Study. Aging and Mental Health, 13, 699–705. doi: 10.1080/13607860902845541 [DOI] [PubMed] [Google Scholar]
  32. Ostir, G. V., Ottenbacher, K. J., & Markides, K. S. (2004). Onset of frailty in older adults and the protective role of positive affect. Psychology and Aging, 19, 402–408. doi: 10.1037/0882-7974.19.3.402 [DOI] [PubMed] [Google Scholar]
  33. Phyo, A. Z. Z., Gonzalez-Chica, D. A., Stocks, N. P., Storey, E., Woods, R. L., Murray, A. M., Orchard, S. G., Shah, R. C., Gasevic, D., Freak-Poli, R., & Ryan, J. (2021). The utility of assessing health-related quality of life to predict cognitive decline and dementia. Journal of Alzheimer's Disease, 80, 895–904. doi: 10.3233/jad-201349 [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Pusswald, G., Tropper, E., Kryspin-Exner, I., Moser, D., Klug, S., Auff, E., Dal-Bianco, P., & Lehrner, J. (2015). Health-related quality of life in patients with subjective cognitive decline and mild cognitive impairment and its relation to activities of daily living. Journal of Alzheimer's Disease, 47, 479–486. doi: 10.3233/jad-150284 [DOI] [PubMed] [Google Scholar]
  35. Robertson, D. A., Savva, G. M., & Kenny, R. A. (2013). Frailty and cognitive impairment—A review of the evidence and causal mechanisms. Ageing Research Reviews, 12, 840–851. doi: 10.1016/j.arr.2013.06.004 [DOI] [PubMed] [Google Scholar]
  36. Rockwood, K., Song, X., MacKnight, C., Bergman, H., Hogan, D. B., McDowell, I., & Mitnitski, A. (2005). A global clinical measure of fitness and frailty in elderly people. Canadian Medical Association Journal, 173, 489–4 95. doi: 10.1503/cmaj.050051 [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Rolfson, D. B., Majumdar, S. R., Tsuyuki, R. T., Tahir, A., & Rockwood, K. (2006). Validity and reliability of the Edmonton Frail Scale. Age and Ageing, 35, 526–529. doi: 10.1093/ageing/afl041 [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Romero-Ortuno, R. (2013). The Frailty Instrument for primary care of the Survey of Health, Ageing and Retirement in Europe predicts mortality similarly to a frailty index based on comprehensive geriatric assessment. Geriatrics & Gerontology International, 13, 497–504. doi: 10.1111/j.1447-0594.2012.00948.x [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Rowe, J. W. & Kahn, R. L. (1998). Successful aging: The MacArthur foundation study. Dell Publishing. [Google Scholar]
  40. Samper-Ternent, R., Al Snih, S., Raji, M. A., Markides, K. S., & Ottenbacher, K. J. (2008). Relationship between frailty and cognitive decline in older Mexican Americans. Journal of the American Geriatrics Society, 56, 1845–1852. doi: 10.1111/j.1532-5415.2008.01947.x [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Santini, Z. I., Jose, P. E., Koyanagi, A., Meilstrup, C., Nielsen, L., Madsen, K. R., & Koushede, V. (2020). Formal social participation protects physical health through enhanced mental health: A longitudinal mediation analysis using three consecutive waves of the Survey of Health, Ageing and Retirement in Europe (SHARE). Social Science and Medicine, 251, 112906. doi: 10.1016/j.socscimed.2020.112906 [DOI] [PubMed] [Google Scholar]
  42. Sha, S., Pan, Y., Xu, Y., & Chen, L. (2022). Associations between loneliness and frailty among older adults: Evidence from the China Health and Retirement Longitudinal Study. BMC Geriatrics, 22, 537. doi: 10.1186/s12877-022-03044-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Siriwardhana, D. D., Weerasinghe, M. C., Rait, G., Scholes, S., & Walters, K. R. (2019). The association between frailty and quality of life among rural community-dwelling older adults in Kegalle district of Sri Lanka: A cross-sectional study. Quality of Life Research, 28, 2057–2068. doi: 10.1007/s11136-019-02137-5 [DOI] [PubMed] [Google Scholar]
  44. Sivertsen, H., Bjørkløf, G. H., Engedal, K., Selbæk, G., & Helvik, A. S. (2015). Depression and quality of life in older persons: A review. Dementia and Geriatric Cognitive Disorders, 40, 311–339. doi: 10.1159/000437299 [DOI] [PubMed] [Google Scholar]
  45. Song, X., Mitnitski, A., & Rockwood, K. (2010). Prevalence and 10-year outcomes of frailty in older adults in relation to deficit accumulation. Journal of the American Geriatrics Society, 58, 681–687. doi: 10.1111/j.1532-5415.2010.02764.x [DOI] [PubMed] [Google Scholar]
  46. Veronese, N., Custodero, C., Cella, A., Demurtas, J., Zora, S., Maggi, S., Barbagallo, M., Sabbà, C., Ferrucci, L., & Pilotto, A. (2021). Prevalence of multidimensional frailty and pre-frailty in older people in different settings: A systematic review and meta-analysis. Ageing Research Reviews, 72, 101498. doi: 10.1016/j.arr.2021.101498 [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Veronese, N., Noale, M., Cella, A., Custodero, C., Smith, L., Barbagelata, M., Maggi, S., Barbagallo, M., Sabbà, C., Ferrucci, L., & Pilotto, A. (2022). Multidimensional frailty and quality of life: Data from the English Longitudinal Study of Ageing. Quality of Life Research, 31, 2985–2993. doi: 10.1007/s11136-022-03152-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. WHOQoL Group. (1993). Study protocol for the World Health Organization project to develop a Quality of Life assessment instrument (WHOQOL). Quality of Life Research, 2, 153–159. doi: 10.1007/BF00435734 [DOI] [PubMed] [Google Scholar]
  49. Xu, H., Yang, X., Lai, X., Zhao, C., Tu, X., Ding, N., Ruan, S., Jiang, Y., Lv, Y., & Zhang, G. (2022). Longitudinal relationships among perceived stress, suicidal ideation and sleep quality in Chinese undergraduates: A cross-lagged model. Journal of Affective Disorders, 299, 45–51. doi: 10.1016/j.jad.2021.11.033 [DOI] [PubMed] [Google Scholar]
  50. Xu, R., Li, Q., Guo, F., Zhao, M., & Zhang, L. (2021). Prevalence and risk factors of frailty among people in rural areas: A systematic review and meta-analysis. BMJ Open, 11, e043494. doi: 10.1136/bmjopen-2020-043494 [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

gbad013_suppl_Supplementary_Tables
Click here for additional data file. (258.5KB, docx)

Data Availability Statement

Because the data are publicly available (https://share-eric.eu/data/) in the Harmonized SHARE data set and Codebook, Version E as of October 2019 developed by the Gateway to Global Aging Data, we do not share data.

Articles from The Journals of Gerontology Series B: Psychological Sciences and Social Sciences are provided here courtesy of Oxford University Press

RESOURCES