Abstract
Background
Few studies have investigated how the joint dynamic evolution of sleep disorders and frailty affects cognitive functions.
Methods
This prospective cohort study used data from the China Health and Retirement Longitudinal Study (CHARLS) from 2011 to 2020, ultimately including 7,749 participants. Sleep duration outside the 6–8 h reference range was used to quantify abnormal sleep duration, with greater departures from this range indicating higher sleep-duration deviation. A group-based trajectory model (GBTM) was used to identify joint longitudinal patterns of sleep deviation risk and frailty, resulting in four distinct joint trajectory groups. Subsequently, Multivariate linear mixed-effects models were then used to evaluate longitudinal changes in continuous cognitive function scores across trajectory groups, while Cox proportional hazards regression models were used to estimate the risk of incident mild cognitive impairment (MCI).
Results
Compared with Group 1 (stable sleep and non-frail), Groups 2 (persistently worsening sleep with pre-frailty), 3 (worsening sleep with progressive pre-frailty), and 4 (persistently worsening sleep with worsening frailty) showed faster cognitive decline (β = −0.017, − 0.022, and − 0.044 SD/year, respectively). Compared with Group 1, Group 4 had the highest risk of incident MCI (HR = 1.48, 95% CI: 1.20–1.82). Elevated risks were also observed in Group 2 (HR = 1.20, 95% CI: 1.04–1.39) and Group 3 (HR = 1.18, 95% CI: 1.02–1.36). Compared with models based on sleep trajectories or frailty trajectories alone, the joint trajectory model yielded a lower AIC and a higher C-index, indicating superior model fit and discriminatory ability for incident MCI.
Conclusions
Worsening joint trajectories of sleep deviation risk and frailty were associated with progressively faster cognitive decline and a higher risk of incident MCI. These joint patterns may provide a useful approach for identifying older adults at elevated cognitive risk.
Supplementary Information
The online version contains supplementary material available at https://doi.org/10.1186/s12889-026-28448-0.
Keywords: Aging, Cognitive decline, Frailty, Sleep duration, Cohort study
Introduction
Mild cognitive impairment (MCI) is a transitional state between normal aging and dementia and has become an increasingly prominent public health concern in the context of global population aging. It is characterized by declines in memory, attention, and executive function, with the basic capacity for activities of daily living preserved [1]. Epidemiological surveys have shown that approximately 15.07 million people aged 60 years or older in China have dementia, whereas about 38.77 million have MCI [2]. Since 2012, the growth rate of the age-standardized incidence of early-onset dementia in China (2.35%) has significantly exceeded the corresponding rate for late-onset dementia (1.15%) [3]. Although individuals with MCI retain a certain degree of functional independence, the large population affected by MCI often requires increased medical attention, monitoring, and support, contributing to rising healthcare and societal costs [4, 5]. Recent evidence suggests that early-onset cognitive impairment has the highest average lifetime cost, approximately US$376,000, with nearly half of this burden occurring during the MCI stage [6].
In recent years, disease-modifying therapies, particularly monoclonal antibodies targeting β-amyloid, have opened new directions for the treatment of Alzheimer’s disease (AD) [7]. However, their clinical application remains constrained by factors such as cost, safety concerns, and patient eligibility criteria [7, 8]. Moving the intervention gateway for AD forward to the MCI stage is one of the core strategies in current AD prevention and treatment. MCI represents an early phase in the Alzheimer’s continuum; in particular, AD-related MCI is regarded as a “golden window” for delaying disease progression [9]. Notably, MCI is not a homogeneous or inevitably progressive condition: 41.5% of patients with MCI progress to dementia within 5 years [10], whereas other studies have shown that approximately 10% to 40% of patients may recover normal cognitive function within 4–5 years [11, 12]. This variability in prognosis suggests that the MCI stage is not merely a precursor to dementia, but also a dynamic phase during which the disease trajectory can still be modified. Therefore, precise identification of high-risk individuals with MCI can facilitate more efficient allocation of clinical resources and timely implementation of pharmacological treatments, behavioral interventions, and multidomain prevention strategies. Furthermore, accumulating evidence indicates that early comprehensive interventions targeting modifiable risk factors help reduce the risk of further cognitive decline [13]. Therefore, focusing on the MCI stage and its high-risk populations represents an important public health strategy for delaying dementia onset and reducing the future burden of dementia.
With social transformation and accelerated urbanization, sleep health has become an emerging public health concern, accompanied by increasing health and economic burdens [14]. Sleep duration is an important indicator of sleep status. Existing studies suggest a nonlinear relationship between sleep duration and cognitive outcomes, with both short and long sleep duration being associated with an increased risk of MCI and disease progression [15, 16]. Meanwhile, sleep duration is closely related to frailty, and abnormal sleep duration and frailty often coexist among middle-aged and older adults [17, 18]. Frailty itself is also an important risk phenotype for MCI [19, 20]. Frailty-related indicators can comprehensively reflect an individual’s overall health status and functional reserve, making them valuable for identifying high-risk populations and prioritizing prevention and intervention for MCI [21]. Importantly, sleep deviation and frailty may not contribute to cognitive impairment as two isolated risk factors. Persistent sleep deviation may be associated with physiological dysregulation, inflammatory and metabolic disturbances, and reduced neurocognitive resilience, whereas frailty reflects decreased functional reserve and increased vulnerability to external and internal stressors [22–24]. Therefore, the coexistence and longitudinal co-evolution of sleep deviation and frailty may indicate an accumulated state of cognitive vulnerability.
Based on this rationale, it is necessary to systematically evaluate the joint longitudinal effect of sleep deviation risk and frailty on cognitive impairment. Although previous studies have linked sleep duration, sleep disturbance, and frailty to cognitive decline, most have examined these factors separately or relied on single-time measurements. Therefore, an important knowledge gap remains regarding whether sleep deviation risk and frailty co-evolve over time, whether their joint longitudinal patterns confer differential risks of cognitive decline and incident MCI, and whether such joint trajectories provide additional risk-stratification value beyond single-exposure trajectories. To address this gap, this study used multi-wave follow-up data from a large national cohort. Nighttime sleep duration was transformed into an index of “sleep deviation risk”, and this index was combined with the frailty index to construct a joint trajectory model. We further examined whether distinct joint trajectory groups were associated with different rates of cognitive decline and risks of incident MCI.
Methods
Study population
This study is prospective, using CHARLS data from 2011 to 2020. CHARLS is an important project supported by the National Natural Science Foundation of China, which mainly collects family and individual information of middle-aged and elderly people aged 45 and above in China [25]. The queue adopts multi-stage probability sampling to ensure national representation, and systematically obtains multi-dimensional data such as demography, health and economic status. CHARLS has been approved by Peking University Biomedical Ethics Committee (IRB00001052-11015), and all subjects have signed the informed consent form.
In this study, baseline was defined using non-overlapping eligible participants from Wave 1 (2011) and Wave 2 (2013). Sleep-duration deviation and frailty index data from Wave 1 to Wave 4 (2011, 2013, 2015, and 2018) were used to construct the joint trajectory model in chronological sequence. Wave 5 (2020) was not included in the trajectory construction and was used only to determine subsequent incident MCI outcomes. Exclusion criteria for participants were: (1) baseline age < 45 years; (2) insufficient sleep duration follow-up data (< 3 waves); (3) insufficient frailty index follow-up data (< 3 waves); (4) baseline diagnosis of MCI or dementia; (5) missing cognitive data or loss to follow-up during the study period. A total of 7,749 participants were ultimately included, with the detailed screening process shown in Supplementary Figure S1.
Sleep deviation risk
Given prior research suggesting a potentially nonlinear relationship between sleep duration and cognitive outcomes [15, 16], nighttime sleep duration was transformed into a measure of deviation from the optimal sleep range of 6–8 h [26, 27] to characterize sleep deviation exposure [28, 29]. Sleep deviation risk was defined as the shortest distance between sleep duration and the interval of 6–8 h. Values within this range were coded as 0; deviations of 1 h, such as 5–9 h, were coded as 1; deviations of 2 h, such as 4–10 h, were coded as 2; and so on. This metric captures both short- and long-sleep deviations on a unified scale while reducing information loss from simple categorization. As this derived variable was nonnegative, integer-valued, zero-heavy, and right-skewed, a zero-inflated Poisson distribution was used in the trajectory model to accommodate its empirical distribution.
Frailty index
In this study, the cumulative defect method is used to construct the vulnerability index (FI), and the FI values of each wave number of Wave 1–4 are calculated. FI consists of 32 health defects, covering such dimensions as coexistence of multiple diseases, feeling and overall health, limited function of daily life and psychological state [30]. According to the preset standard, all indicators are recoded as existence (= 1) or nonexistence (= 0). Among them, visual impairment and self-rated health come from the re-coding of questionnaire items; Depression symptoms were classified according to CESD-10 (0–10 = 0; 11–30 = 1); Cognitive deficit is calculated by combining memory recall and directional performance, and standardized to 0–1. The total number of defects is obtained by adding 32 defects. In order to ensure the consistency of measurement, if any item of a wave is missing, the wave FI is recorded as missing. FI is expressed on a scale of 0–100, and the calculation formula is: FI = (number of deficits / 32) × 100. See Supplementary Table S1 for the specific definitions and recoding rules of each index, and Supplementary Figures S2–S3 for the distribution and age change characteristics of baseline FI.
Cognitive decline
In CHARLS, cognitive function was assessed biennially across episodic memory, orientation, and executive function [31]. Episodic memory was measured using immediate and delayed recall of 10 words. Orientation was assessed using five time-related items (month, day, year, weekday, and season; score range: 0–5). Executive function was evaluated using the Serial 7s test, selected orientation items, and a figure-copying task, with scores summed to yield a total ranging from 0 to 11.
Cognitive Z-scores were constructed to standardize measures across domains with different score ranges [32]. Baseline means and standard deviations were used to standardize domain-specific scores from Waves 1–5. Global cognition was defined as the average of the three domain Z-scores and further standardized using baseline parameters. A lower cognitive Z-score indicates worse cognitive performance relative to the baseline population mean. Therefore, greater declines in Z-scores over time were interpreted as reflecting faster cognitive decline in this study.
Following the AACD criteria [33, 34], MCI was defined as performance at least 1 standard deviation below the mean of the corresponding age group (stratified in 5-year bands).
Trajectory recognition and naming
Participants were included in the primary analysis if they had concurrent observations of sleep deviation risk and FI in at least three follow-up waves. We applied group-based multi-trajectory modeling (GBMT) to jointly model the longitudinal trajectories of sleep deviation risk and FI, specifying a zero-inflated Poisson distribution for sleep deviation risk and a censored normal distribution for FI [35, 36]. Models with varying numbers of latent classes (K = 2–6) were fitted and compared using the Bayesian information criterion (BIC), entropy, mean posterior probability, and class proportions to avoid overly small groups. The optimal solution was selected based on the lowest BIC, a mean posterior probability ≥ 70%, entropy ≥ 0.70, and a minimum class size of ≥ 5%, yielding a four-class model [36]. Specifically, Group 1 (G1) was defined as “Stable sleep and non-frailty”, Group 2 (G2) as “Persistently worsening sleep with pre-frailty”, Group 3 (G3) as “Worsening sleep with progressive pre-frailty”, and Group 4 (G4) as “Persistently worsening sleep with worsening frailty”. To compare the prognostic performance of joint versus single-domain trajectories, sleep trajectory groups (ST) and frailty trajectory groups (FT) were additionally derived using group-based trajectory models for each domain separately and evaluated in Cox models. The detailed model-fitting and selection process is reported in the Supplementary methods and Tables S2–S7.
Covariates
We included the following baseline covariates as potential confounders: age; sex (male or female); marital status (married or other); education level (below primary, primary, junior secondary, or senior secondary and above); residence (urban or rural); smoking status (ever smoker or never smoker); drinking status (ever drinker or never drinker); and social isolation. Social isolation was assessed using a 0–4 index, computed as the sum of four binary items: being unmarried, living alone, having contact with children less than once per week, and not participating in any social activities in the past month [37].
Statistical analysis
Baseline characteristics were described for the joint trajectory groups, with categorical variables reported as counts (percentages) and continuous variables as means (standard deviations) or medians (interquartile ranges), as appropriate. Chi-square test was used for classified variables, and one-way ANOVA or Kruskal-Wallis rank sum test was used for continuous variables.
To evaluate longitudinal changes in cognition during follow-up, we fitted linear mixed-effects models (LMMs) to repeated cognitive measurements [36]. LMMs account for within-person correlation, accommodate incomplete repeated measurements under a missing-at-random assumption, and allow individual variation in baseline cognition and change over time. The multitrajectory group was the primary exposure, with G1 as the reference. Models included follow-up time, trajectory group, and their interaction to compare rates of cognitive change across groups, with individual-level random intercepts and random slopes for time.
Associations between multitrajectory groups and incident MCI were evaluated with Cox proportional hazards models, reporting hazard ratios (HRs) and 95% confidence intervals (CIs). Person-time accrued from baseline to MCI diagnosis, death, loss to follow-up, or the end of follow-up, whichever occurred first. The proportional hazards assumption was assessed using Schoenfeld residuals, and no violations were detected (Supplementary Table S8).
Covariates were adjusted for in a stepwise manner: Model 1 adjusted for age and sex, and Model 2 further adjusted for marital status, education level, location, smoking, drinking, and social isolation. Prespecified subgroup analyses were conducted to assess whether associations differed across population strata by including interaction terms between trajectory group and subgroup variables. To examine the robustness of our findings, we performed several sensitivity analyses, including (1) additionally adjusting for baseline cognitive score in the mixed-effects models, (2) using an alternative composite cognitive score derived as the sum of memory, orientation, and executive function measures, (3) repeating the analyses after excluding cognition-related frailty components, and (4) applying inverse probability of censoring weighting (IPCW) to address potential informative loss to follow-up, as participants at higher risk of cognitive decline may have been more likely to drop out. IPCW was used under the assumption that censoring was conditionally independent of future outcomes given observed covariates.
All tests were two-sided, and p < 0.05 was considered statistically significant. Joint trajectory models were fit in Stata (version 17.0) using the “traj” procedure. Subsequent analyses were conducted in R (version 4.3.3), and linear mixed-effects models were implemented using the “lme4” package.
Results
Baseline characteristics
Figure 1 shows the longitudinal trajectory of the frailty index and the risk of sleep deviation during the follow-up period. The joint trajectory can be divided into four groups: G1 (n = 4,178; 53.92%), G2 (n = 1,406; 18.14%), G3 (n = 1,675; 21.26%), and G4 (n = 490; 6.32%). G1 was characterized by stable sleep and non-frailty, representing the most favorable joint trajectory. G2 was characterized by persistently worsening sleep with pre-frailty, reflecting sustained sleep-duration deviation with relatively low-to-moderate frailty. G3 showed worsening sleep with progressive pre-frailty, indicating a gradual increase in frailty alongside sleep-duration deviation. G4 was characterized by persistently worsening sleep with worsening frailty and represented the most unfavorable overall joint trajectory, showing the highest frailty level and the steepest increase in frailty over follow-up.
Fig. 1.
Trajectories of sleep deviation risk and frailty from Wave 1 to Wave 4 (2011–2018). G1: Stable sleep and non-frail; G2: Persistently worsening sleep with pre-frailty; G3: Worsening sleep with progressive pre-frailty; G4: Persistently worsening sleep with worsening frailty
Table 1 reports the baseline characteristics of 7,749 participants. There were significant differences in all baseline variables among the four groups. From G1 to G4, the age of participants gradually increased (55.05 ± 7.78 to 61.63 ± 8.11 years old), the proportion of women increased (39.85% to 63.88%), the proportion of unmarried people increased (5.48% to 14.90%), and the education level decreased (high school and above: 20.01% to 5.92%). In terms of lifestyle, compared with G1, the proportion of smokers and drinkers in G2–G4 group is lower, while social isolation is more obvious. The sleep duration was also different among the groups, with G1 being the highest (6.95 ± 1.16 h) and G4 being the lowest (5.19 ± 2.13 h). At the same time, the debilitating burden increased significantly from G1(7.31 ± 4.84) to G4(28.03 ± 11.72), suggesting that the high-risk group has a higher debilitating level.
Table 1.
Baseline characteristics of the multitrajectory groups by sleep deviation risk and frailty
| Characteristic | Total | G1 | G2 | G3 | G4 | P |
|---|---|---|---|---|---|---|
| Number | 7749 | 4178 | 1406 | 1675 | 490 | |
| Age, years* | 56.73(8.31) | 55.05(7.78) | 57.42(8.55) | 58.93(8.31) | 61.63(8.11) | < 0.001 |
| Sex, % | < 0.001 | |||||
| Male | 4127(53.26) | 2513(60.15) | 698(49.64) | 739(44.12) | 177(36.12) | |
| Female | 3622(46.74) | 1665(39.85) | 708(50.36) | 936(55.88) | 313(63.88) | |
| Marital status, % | < 0.001 | |||||
| Marital | 7149(92.26) | 3949(94.52) | 1288(91.61) | 1495(89.25) | 417(85.10) | |
| Others | 600(7.74) | 229(5.48) | 118(8.39) | 180(10.75) | 73(14.90) | |
| Education, % | < 0.001 | |||||
| Below primary | 2561(33.05) | 1060(25.37) | 556(39.54) | 673(40.18) | 272(55.51) | |
| Primary | 1929(24.89) | 975(23.34) | 375(26.67) | 466(27.82) | 113(23.06) | |
| Junior secondary | 2059(26.57) | 1307(31.28) | 310(22.05) | 366(21.85) | 76(15.51) | |
| Senior secondary or higher | 1200(15.49) | 836(20.01) | 165(11.74) | 170(10.15) | 29(5.92) | |
| Location, % | < 0.001 | |||||
| Urban | 3111(40.15) | 1804(43.18) | 531(37.77) | 616(36.78) | 160(32.65) | |
| Rural | 4638(59.85) | 2374(56.82) | 875(62.23) | 1059(63.22) | 330(67.35) | |
| Smoking status, % | < 0.001 | |||||
| Ever smoker | 3258(42.06) | 1745(45.44) | 720(38.34) | 680(39.28) | 135(33.47) | |
| Never smoker | 4488(57.94) | 2123(54.56) | 1090(61.66) | 1139(60.72) | 248(66.53) | |
| Drinking status, % | < 0.001 | |||||
| Ever drinker | 3486(45.02) | 2072(49.64) | 585(41.61) | 651(38.87) | 178(36.40) | |
| Never drinker | 4258(54.98) | 2102(50.36) | 821(58.39) | 1024(61.13) | 311(63.60) | |
| Social isolation* | 0.64(0.69) | 0.56(0.65) | 0.67(0.69) | 0.73(0.73) | 0.89(0.78) | < 0.001 |
| Sleep duration, hours* | 6.45(1.70) | 6.95(1.16) | 5.49(2.26) | 6.38(1.55) | 5.19(2.13) | < 0.001 |
| Frailty index* | 11.42(8.75) | 7.31(4.84) | 10.25(6.07) | 18.50(7.82) | 28.03(11.72) | < 0.001 |
*Data are expressed as mean (SD)
Abbreviations: G1 Stable sleep and non-frail, G2 Persistently worsening sleep with pre-frailty, G3 Worsening sleep with progressive pre-frailty, G4 Persistently worsening sleep with worsening frailty
Multitrajectories of sleep deviation risk and frailty with cognitive decline
The detailed results are shown in Table 2, supplementary Figure S4 shows the cognitive change trajectory of different risk groups of frailty-sleep deviation. With G1 as reference, both models showed that cognitive function decreased significantly with time (Time: β ≈ −0.035 per year; p < 0.001). At baseline, the cognitive scores of G2, G3 and G4 were significantly lower than that of G1, and this difference still existed after full adjustment (Model 2: G2 β=−0.148, G3 β=−0.121, G4 β=−0.266; all p < 0.001). At the same time, the interaction term between group and time is negative and statistically significant, indicating that the cognitive decline rate of G2, G3 and G4 is faster than that of G1, and G4 is the most obvious (Model 2: G2×time β=−0.017, G3×time β=−0.022, G4×time β=−0.044; all p < 0.001). Generally speaking, the trajectory group with higher risk showed lower baseline cognitive level and faster cognitive decline during follow-up, showing clear gradient characteristics.
Table 2.
Fixed-effect estimates from linear mixed-effects models for cognitive decline by multitrajectories of sleep deviation risk and frailty
| Model 1 | Model 2 | |||
|---|---|---|---|---|
| β(95%CI) | P | β(95%CI) | P | |
| Time, years | −0.034(−0.038,−0.031) | < 0.001 | −0.035(−0.038,−0.031) | < 0.001 |
| Multitrajectories of sleep deviation risk and frailty | ||||
| G1 | Reference | Reference | ||
| G2 | −0.241(−0.278,−0.203) | < 0.001 | −0.148(−0.180,−0.116) | < 0.001 |
| G3 | −0.213(−0.249,−0.178) | < 0.001 | −0.121(−0.151,−0.090) | < 0.001 |
| G4 | −0.441(−0.500,−0.382) | < 0.001 | −0.266(−0.317,−0.215) | < 0.001 |
| Multitrajectories of sleep deviation risk and frailty × time | ||||
| G1×time | Reference | Reference | ||
| G1×time | −0.017(−0.025,−0.010) | < 0.001 | −0.017(−0.025,−0.010) | < 0.001 |
| G2×time | −0.021(−0.029,−0.014) | < 0.001 | −0.022(−0.029,−0.014) | < 0.001 |
| G4×time | −0.044(−0.056,−0.032) | < 0.001 | −0.044(−0.056,−0.032) | < 0.001 |
Model 1 was adjusted for baseline age and sex. Model 2 was adjusted for baseline age, sex, marital status, educational level, location, smoking status, drinking status and social isolation
Abbreviations: G1 Stable sleep and non-frail, G2 Persistently worsening sleep with pre-frailty, G3 Worsening sleep with progressive pre-frailty, G4 Persistently worsening sleep with worsening frailty
Multitrajectories of sleep deviation risk and frailty with MCI
In the fully adjusted model, sleep deviation risk alone, frailty alone, and their joint trajectories were all independently associated with an increased risk of MCI compared with the stable or low-risk groups. Among the joint trajectory groups, Group 4 showed the highest risk (Table 3). Under the same covariate-adjustment framework, the joint trajectory model retained the lowest AIC value and the highest C-index, and the likelihood-ratio test remained statistically significant compared with the base model. These results indicate that the joint sleep–frailty trajectory provides additional predictive value beyond individual trajectories. In addition, we did not observe a significant multiplicative interaction between sleep trajectories and frailty trajectories (Supplementary Tables S9 − S10).
Table 3.
MCI incidence and risk estimates across joint and separate trajectories of sleep deviation risk and frailty
| Group | Case | Case/PY | Model1 | Model2 |
|---|---|---|---|---|
| Trajectory of sleep deviation risk | ||||
| ST1 | 857 | 857/5558 | Reference | Reference |
| ST2 | 442 | 442/2191 | 1.38 (1.23 − 1.55) *** | 1.16 (1.03 − 1.30) * |
| Trajectory of frailty | ||||
| FT1 | 811 | 811/5234 | Reference | Reference |
| FT2 | 382 | 382/2075 | 1.28 (1.13 − 1.45) *** | 1.08 (0.95 − 1.23) |
| FT3 | 106 | 106/440 | 1.89 (1.54 − 2.33) *** | 1.52 (1.23 − 1.87) *** |
| Multitrajectories of sleep deviation risk and frailty | ||||
| G1 | 600 | 600/33,022 | Reference | Reference |
| G2 | 274 | 274/11,173 | 1.42 (1.23 − 1.65) *** | 1.20 (1.04 − 1.39) * |
| G3 | 312 | 312/13,325 | 1.42 (1.23 − 1.63) *** | 1.18 (1.02 − 1.36) * |
| G4 | 113 | 113/3878 | 1.98 (1.61 − 2.44) *** | 1.48 (1.20 − 1.82) *** |
Abbreviations: PY Person-years, ST1 Stable sleep trajectory, ST2 Worsening sleep trajectory, FT1 Non-frail trajectory, FT2 Pre-frail trajectory, FT3 Progressive frailty trajectory, G1 Stable sleep and non-frail, G2 Persistently worsening sleep with pre-frailty, G3 Worsening sleep with progressive pre-frailty, G4 Persistently worsening sleep with worsening frailty
*P < 0.05; ***P < 0.001
Multitrajectories of sleep deviation risk and frailty with subdomains of cognitive decline
Table 4 presents detailed results. Supplementary Figure S5 shows the predicted trajectories of domain-specific cognitive z-scores across multitrajectory groups of sleep deviation risk and frailty. Using G1 as the reference, participants in G2–G4 had generally lower baseline performance across the three cognitive domains (negative β estimates; most p < 0.001), with the largest deficits observed in G4. Over follow-up, orientation and executive function declined significantly with time, whereas the overall time effect for memory was not significant. However, compared with G1, G2-G4’ s memory function declines faster (all group-by-time interactions p < 0.001), and G4’ s directional ability and executive function decline more significantly, indicating that the higher-risk trajectory group is often accompanied by more extensive and faster cognitive decline.
Table 4.
Domain-specific cognitive outcomes by multitrajectories of sleep deviation risk and frailty
| Memory | Orientation | Executive function | ||||
|---|---|---|---|---|---|---|
| β(95%CI) | P | β(95%CI) | P | β(95%CI) | P | |
| Time, years | −0.003(−0.008,0.003) | 0.356 | −0.032(−0.037,−0.028) | < 0.001 | −0.047(−0.050,−0.040) | < 0.001 |
| Multitrajectories of sleep deviation risk and frailty | ||||||
| G1 | Reference | Reference | Reference | |||
| G2 | −0.119(−0.158,−0.080) | < 0.001 | −0.141(−0.173,−0.109) | < 0.001 | −0.097(−0.131,−0.063) | < 0.001 |
| G3 | −0.139(−0.176,−0.101) | < 0.001 | −0.114(−0.145,−0.083) | < 0.001 | −0.037(−0.069,−0.004) | 0.027 |
| G4 | −0.250(−0.312,−0.187) | < 0.001 | −0.243(−0.294,−0.192) | < 0.001 | −0.138(−0.192,−0.083) | < 0.001 |
| Multitrajectories of sleep deviation risk and frailty × time | ||||||
| G1×time | Reference | Reference | Reference | |||
| G2×time | −0.028(−0.039,−0.017) | < 0.001 | −0.006(−0.015,0.003) | 0.191 | −0.011(−0.020,−0.003) | 0.013 |
| G3×time | −0.030(−0.040,−0.019) | < 0.001 | −0.010(−0.018,−0.001) | 0.019 | −0.008(−0.017,−0.001) | 0.048 |
| G4×time | −0.047(−0.064,−0.030) | < 0.001 | −0.034(−0.047,−0.020) | < 0.001 | −0.020(−0.034,−0.006) | < 0.01 |
The model was adjusted for baseline age, sex, marital status, educational level, location, smoking status, drinking status and social isolation
Abbreviations: G1 Stable sleep and non-frail, G2 Persistently worsening sleep with pre-frailty, G3 Worsening sleep with progressive pre-frailty, G4 Persistently worsening sleep with worsening frailty
Subgroup analysis
To examine the robustness of our findings across population subgroups and assess potential heterogeneity, we performed prespecified subgroup analyses comparing G4 versus G1. Overall, the association between G4 membership and higher risk of incident MCI was broadly consistent across subgroups, with no significant effect modification observed (all P for interaction > 0.05). In subgroup analyses using LMM for cognitive decline, the pattern of faster annual decline in G4 relative to G1 remained generally stable across most strata; However, differential effects were observed among older participants and those with lower educational attainment, particularly for executive function and global cognitive decline. In addition, interaction signals were observed for marital status, residence, and smoking status in models for memory and/or orientation. Detailed subgroup-specific estimates and interaction tests are provided in the Supplementary Figures S6 − S10.
Sensitivity analysis
In sensitivity analyses, we evaluated the robustness of our findings from four perspectives. First, to assess whether the results were dependent on the definition of cognitive outcome, the composite cognitive score was re-estimated as the sum of memory, orientation, and executive function measures. Second, to examine whether baseline cognitive status influenced the observed longitudinal associations, models were further adjusted for baseline cognitive score. Third, because the frailty index included cognitive-status items, which could conceptually and methodologically overlap with cognitive outcomes, we repeated the analyses after excluding cognition-related frailty components. Fourth, to reduce potential bias from informative loss to follow-up, inverse probability of censoring weighting was applied. The main results were materially unchanged across these analyses (Supplementary Tables S11–S15).
Discussion
Principal findings
In this large-scale, nationwide prospective cohort study, we constructed distinct joint trajectory groups based on sleep deviation risk and frailty, and found that these trajectories were significantly associated with both the rate of cognitive decline and the risk of incident MCI. Compared with participants with stable sleep and no frailty (G1), those in the higher-risk trajectory groups (G2–G4) showed lower baseline cognitive performance, faster cognitive decline during follow-up, and a higher risk of incident MCI, with the most pronounced associations observed in G4, which was characterized by persistently worsening sleep deviation risk and worsening frailty. The joint trajectory model showed better predictive performance than the single-exposure trajectory models. In domain-specific analyses, G2–G4 showed faster decline in episodic memory, whereas accelerated deterioration in orientation and executive function was mainly concentrated in the highest-risk group, especially G4. Subgroup and sensitivity analyses further supported the robustness of the main results.
Comparison with previous studies
In the present study, the most pronounced cognitive decline and the highest risk of incident MCI were observed in G4, which was characterized by persistently worsening sleep deviation risk and worsening frailty. This finding is consistent with previous studies showing that both short and long sleep duration are associated with an increased risk of cognitive impairment. Specifically, compared with older adults sleeping 6–8 h per night, those sleeping more than 8 h per night were more likely to develop MCI and showed poorer performance in global cognition, verbal fluency, attention, and executive function [15, 26]. Similarly, short sleep duration (< 6 h) has been associated with a 22%–24% increased risk of MCI, and extremely short sleep duration (≤ 3 h) may shorten the average time from MCI to dementia by approximately half [16, 38].
Frailty is also an increasing global health burden and generally reflects a decline in an individual’s ability to cope with internal and external stressors [24]. With population aging, the prevalence of frailty continues to rise. Epidemiological evidence has shown that frailty is closely associated with future cognitive decline and increased MCI risk, and this association tends to become stronger with increasing frailty severity [19, 20]. This is in line with our findings. Conversely, MCI may also increase the risk of frailty, suggesting that cognitive impairment and frailty may form a mutually reinforcing cycle during age-related functional decline [39]. The coexistence of frailty and MCI is often referred to as “cognitive frailty” [40], which has been associated with a 4–6-fold increased risk of dementia within 4 years and a 2.6-fold increased risk of death within 8 years [41].
It should also be emphasized that sleep deviation and frailty are closely interrelated. A meta-analysis showed that sleeping less than 7 h or more than 8 h per night was associated with a higher risk of frailty [17]. Conversely, frail individuals are more likely to experience insomnia, shortened sleep duration, and sleep apnea [18]. These two conditions may further aggravate each other, thereby increasing vulnerability to cognitive impairment [42, 43]. Although no significant multiplicative interaction between sleep and frailty trajectories was observed, our findings still indicate that their co-evolution provides important risk stratification value. Therefore, the joint trajectory patterns, rather than either factor alone, may help identify older adults at particularly high risk of cognitive decline and incident MCI.
Age should be considered when interpreting these findings. Participants in the higher-risk trajectory groups were older at baseline, which is consistent with age-related disruption in sleep duration and accumulation of frailty [44, 45], suggesting that aging may partly contribute to the clustering of sleep deviation risk, frailty accumulation, and cognitive vulnerability. However, the associations remained significant after adjustment for age and other baseline covariates, indicating that they were not solely explained by age differences among groups. The significant negative group × time interaction further suggests that G2–G4 experienced faster cognitive decline than G1, with the most pronounced decline observed in G4. Together, these findings indicate that joint sleep–frailty trajectories may provide additional information for identifying accelerated cognitive decline beyond chronological age alone.
Potential mechanisms
Several potential mechanisms may explain the observed associations. Chronic sleep disruption may accelerate AD-related pathological processes, while disturbed sleep may also weaken glymphatic clearance by disrupting norepinephrine-mediated slow vasomotion during sleep, thereby reducing the clearance of neurotoxic metabolites [46, 47]. Moreover, sleep disturbances can further contribute to frailty by exacerbating metabolic and inflammatory dysregulation, as well as by altering the composition of gut microbial metabolites [48, 49]. Frailty may also be linked to cognitive impairment through AD-related pathology, neuroinflammation, and neurodegenerative processes. Recent evidence from Chinese older adults suggests that frailty is associated with plasma AD-related biomarkers, including p-tau217, GFAP, NfL, and Aβ42/Aβ40, while inflammatory pathways may further link physical frailty with cognitive impairment [19, 50]. In addition, shared biological and clinical factors, such as cardiovascular dysfunction, aggravated depressive symptoms, decreased testosterone levels, impaired renal function, and increased proinflammatory cytokines, may contribute to both sleep disturbance and frailty [51, 52]. These shared pathways may partly explain why the joint trajectory of worsening sleep deviation risk and frailty accumulation was associated with accelerated cognitive decline and increased MCI risk in our study.
Clinical and public health implications
Accordingly, in clinical and community settings, individuals with both persistent sleep deviation and progressive frailty may represent a priority group for early screening and preventive intervention. Based on our findings, intervention strategies should shift from single-factor management toward integrated sleep–frailty management. For sleep-related problems, non-pharmacological approaches such as cognitive behavioral therapy for insomnia and assessment of treatable sleep disorders may be considered [53]. Improving sleep quality and sleep efficiency may reduce daytime fatigue, enhance physical activity, and increase older adults’ capacity to participate in exercise- and nutrition-based frailty interventions [54]. For frailty, multicomponent interventions including resistance and balance exercise, physical activity promotion, nutritional support, and management of comorbidities may help preserve functional reserve and delay frailty progression [55]. Future intervention studies are warranted to test whether combined sleep improvement and frailty management can delay cognitive decline and reduce MCI risk.
Strengths and limitations
Although this study was based on a large-scale national cohort and adopted a relatively complete longitudinal design, some limitations remain. First, both the frailty index and the night sleep duration were self-reported, which may have introduced reporting or recall bias. Moreover, sleep assessment was based mainly on nighttime sleep duration and did not fully capture other sleep dimensions, such as sleep quality, daytime napping, insomnia symptoms, or sleep-disordered breathing. Second, building a dynamic trajectory based on only four follow-up visits may not fully capture short-term fluctuations or complex nonlinear changes, thus simplifying the trajectory model. Third, as an observational cohort study, this study cannot establish causal relationships, and its generalizability may be limited. Future studies could use Mendelian randomization to explore potential causal relationships and further validate these findings in cross-national aging cohorts. Finally, although many potential confounding factors were controlled, unmeasured factors such as heredity and blood biomarkers may still lead to residual confounding. Future research should combine neuroimaging, blood biomarkers and genetic data to further clarify the biological mechanism between sleep disorder, frailty and cognitive decline.
Conclusion
The results of this study showed that the high-risk joint trajectory group of sleep deviation risk and frailty was significantly associated with cognitive decline and an increased risk of MCI, with faster declines in memory, orientation, and executive function. Moreover, the joint trajectory provided additional predictive value beyond separate sleep deviation risk or frailty trajectories. These findings suggest that individuals at high risk for both sleep disturbance and frailty should be identified and targeted for early intervention to delay cognitive decline.
Supplementary Information
Acknowledgements
We thank the China Health and Retirement Longitudinal Study team for providing the data and the participants for their valuable contributions.
Authors’ contributions
Linfeng Chen: Data curation, Formal analysis, Investigation, Methodology, Validation, Visualization, Writing–original draft; Chuncao Ao: Data curation, Formal analysis, Investigation, Methodology, Validation, Visualization, Writing–original draft; Su Zhang: Investigation, Methodology, Writing–review & editing; Xinhui Zhang: Investigation, Resources, Writing–review & editing; Zhong Li: Conceptualization, Supervision, Project administration, Funding acquisition, Writing–review & editing.
Funding
This study was financially supported by the Science and Technology Planning Project of Guangdong Province, China (No. 2023B1212060018), and the Key Project of Guangzhou Science and Technology Program, China (No. 202206010142).
Data availability
The data underlying this study are available from the China Health and Retirement Longitudinal Study (CHARLS) database. Researchers may obtain access by registering and applying through the official CHARLS website at http://charls.pku.edu.cn.
Declarations
Ethics approval and consent to participate
All methods were conducted in accordance with the Declaration of Helsinki. The China Health and Retirement Longitudinal Study (CHARLS) was approved by the Biomedical Ethics Committee of Peking University (IRB00001052-11015), and written informed consent was obtained from all participants. Clinical trial number: not applicable.
Consent for publication
Not applicable.
Competing interests
The authors declare no competing interests.
Footnotes
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Linfeng Chen and Chuncao Ao contributed equally to this work as co-first authors for this article
References
- 1.Petersen RC. Mild cognitive impairment as a diagnostic entity. J Intern Med. 2004;256(3):183–94. [DOI] [PubMed] [Google Scholar]
- 2.Jia L, Du Y, Chu L, et al. Prevalence, risk factors, and management of dementia and mild cognitive impairment in adults aged 60 years or older in China: a cross-sectional study. Lancet Public Health. 2020;5(12):e661–71. [DOI] [PubMed] [Google Scholar]
- 3.Lu KQ, Lu KJ, Ji ZJ. Comparative analysis of the burden of young-onset and late-onset dementia in China from 1990 to 2021: A study based on GBD 2021 data. J Prev Alzheimers Dis. 2025;12(9):100307. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Connors MH, Seeher K, Teixeira-Pinto A, Woodward M, Ames D, Brodaty H. Mild Cognitive Impairment and Caregiver Burden: A 3-Year-Longitudinal Study. Am J Geriatr Psychiatry. 2019;27(11):1206–15. [DOI] [PubMed] [Google Scholar]
- 5.Frech FH, Li G, Juday TR, et al. Real-world health care costs and resource utilization associated with mild cognitive impairment in the United States: A retrospective cohort study of commercial and Medicare data. J Manag Care Spec Pharm. 2025;31(8):782–94. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Gracner T, Chaturvedi R, Nguyen P, et al. The burden of cognitive impairment. Alzheimers Dement. 2025;21(8):e70436. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Rafii MS, Aisen PS. Amyloid-lowering immunotherapies for Alzheimer disease: current status and future directions. Nat Rev Neurol. 2025;21(9):490–8. [DOI] [PubMed] [Google Scholar]
- 8.Frisoni GB, Aho E, Brayne C, et al. Alzheimer’s disease outlook: controversies and future directions. Lancet. 2025;406(10510):1424–42. [DOI] [PubMed] [Google Scholar]
- 9.Sperling R, Mormino E, Johnson K. The evolution of preclinical Alzheimer’s disease: implications for prevention trials. Neuron. 2014;84(3):608–22. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Salemme S, Lombardo FL, Lacorte E, et al. The prognosis of mild cognitive impairment: A systematic review and meta-analysis. Alzheimers Dement (Amst). 2025;17(1):e70074. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Sanz-Blasco R, Ruiz-Sánchez de León JM, Ávila-Villanueva M, Valentí-Soler M, Gómez-Ramírez J, Fernández-Blázquez MA. Transition from mild cognitive impairment to normal cognition: Determining the predictors of reversion with multi-state Markov models. Alzheimers Dement. 2022;18(6):1177–85. [DOI] [PubMed] [Google Scholar]
- 12.Larrieu S, Letenneur L, Orgogozo JM, et al. Incidence and outcome of mild cognitive impairment in a population-based prospective cohort. Neurology. 2002;59(10):1594–9. [DOI] [PubMed] [Google Scholar]
- 13.Mian M, Tahiri J, Eldin R, Altabaa M, Sehar U, Reddy PH. Overlooked cases of mild cognitive impairment: Implications to early Alzheimer’s disease. Ageing Res Rev. 2024;98:102335. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Liu XX, Wang Z, Chen SJ, et al. Sleep health in China: status, challenges, and promotion strategies. Lancet Public Health. 2025;10(12):e1055–65. [DOI] [PubMed] [Google Scholar]
- 15.Tang S, Liu R, Ren J, et al. Association of objective sleep duration with cognition and brain aging biomarkers in older adults. Brain Commun. 2024;6(3):fcae144. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Song Y, Liu H, Gu K, Liu Y. Association between nighttime sleep duration, nap time, and mild cognitive impairment in Chinese older adults: a cross-sectional study. BMC Public Health. 2024;24(1):2381. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Liu HM, Xue YJ, Tang KW, et al. Association between sleep duration and frailty in older adults: Systematic review and meta-analysis of observational studies. Arch Gerontol Geriatr. 2025;137:105949. [DOI] [PubMed] [Google Scholar]
- 18.Ribeiro BC, de Athayde Costa E, Silva A, de Souza LBR, de Araújo Moraes JB, Carneiro SR, Neves LMT. Risk stratification for frailty, impairment and assessment of sleep disorders in community-dwelling older adults. Exp Gerontol. 2024;187:112370. [DOI] [PubMed] [Google Scholar]
- 19.Tian X, Zhu M, Ma Y, et al. Physical and biopsychosocial frailty, cognitive phenotypes, and plasma biomarkers for Alzheimer’s disease in Chinese older adults: A population-based study. Alzheimers Dement. 2025;21(5):e70303. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Ward DD, Wallace LMK, Rockwood K. Frailty and Risk of Dementia in Mild Cognitive Impairment Subtypes. Ann Neurol. 2021;89(6):1221–5. [DOI] [PubMed] [Google Scholar]
- 21.Wang S, Li Q, Wang S, et al. Sustained frailty remission and dementia risk in older adults: A longitudinal study. Alzheimers Dement. 2024;20(9):6268–77. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Irwin MR, Olmstead R, Carroll JE. Sleep Disturbance, Sleep Duration, and Inflammation: A Systematic Review and Meta-Analysis of Cohort Studies and Experimental Sleep Deprivation. Biol Psychiatry. 2016;80(1):40–52. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Duan D, Kim LJ, Jun JC, Polotsky VY. Connecting insufficient sleep and insomnia with metabolic dysfunction. Ann N Y Acad Sci. 2023;1519(1):94–117. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Clegg A, Young J, Iliffe S, Rikkert MO, Rockwood K. Frailty in elderly people. Lancet. 2013;381(9868):752–62. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Zhao Y, Hu Y, Smith JP, Strauss J, Yang G. Cohort profile: the China Health and Retirement Longitudinal Study (CHARLS). Int J Epidemiol. 2014;43(1):61–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Consortium MULTI, O’Toole CK, Song Z et al. Sleep chart of biological ageing clocks in middle and late life. Nat Published online May 13, 2026. [DOI] [PMC free article] [PubMed]
- 27.Cappuccio FP, Cooper D, D’Elia L, Strazzullo P, Miller MA. Sleep duration predicts cardiovascular outcomes: a systematic review and meta-analysis of prospective studies. Eur Heart J. 2011;32(12):1484–92. [DOI] [PubMed] [Google Scholar]
- 28.Balter LJ, Malmros J, Stenkrona P, et al. A [11 C]PBR28 PET study on the associations between sleep health and microglial density. J Neuroinflammation. 2025;22(1):270. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Attaullah H, Sanaullah S, Peters A, Ahmed QA, Baudisch J, Jungeblut T. DNA: detecting early signs of neurodegenerative diseases through activity and sleep analysis. Front Neurosci. 2025;19:1617758. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Chamberlain AM, Finney Rutten LJ, Manemann SM, et al. Frailty Trajectories in an Elderly Population-Based Cohort. J Am Geriatr Soc. 2016;64(2):285–92. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Rong H, Lai X, Jing R, Wang X, Fang H, Mahmoudi E. Association of Sensory Impairments With Cognitive Decline and Depression Among Older Adults in China. JAMA Netw Open. 2020;3(9):e2014186. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.González HM, Tarraf W, Harrison K, et al. Midlife cardiovascular health and 20-year cognitive decline: Atherosclerosis Risk in Communities Study results. Alzheimers Dement. 2018;14(5):579–89. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Richards M, Touchon J, Ledesert B, Richie K. Cognitive decline in ageing: are AAMI and AACD distinct entities? Int J Geriatr Psychiatry. 1999;14(7):534–40. [DOI] [PubMed] [Google Scholar]
- 34.Hu Y, Peng W, Ren R, Wang Y, Wang G. Sarcopenia and mild cognitive impairment among elderly adults: The first longitudinal evidence from CHARLS. J Cachexia Sarcopenia Muscle. 2022;13(6):2944–52. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Nagin DS, Jones BL, Elmer J. Recent Advances in Group-Based Trajectory Modeling for Clinical Research. Annu Rev Clin Psychol. 2024;20(1):285–305. [DOI] [PubMed] [Google Scholar]
- 36.Hu C, Sun X, Li Z, et al. Multitrajectories of Frailty and Depression With Cognitive Function: Findings From the Health and Retirement Longitudinal Study. J Cachexia Sarcopenia Muscle. 2025;16(2):e13795. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Mehrabi F, Pomeroy ML, Cudjoe TKM, Jenkins E, Dent E, Hoogendijk EO. The temporal sequence and reciprocal relationships of frailty, social isolation and loneliness in older adults across 21 years. Age Ageing. 2024;53(10):afae215. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Ye L, Johnson J, Zhang N. Sleep deficiencies and the speed of cognitive decline in individuals with mild cognitive impairment. Sleep. 2025;48(Suppl 1):A546–7. [Google Scholar]
- 39.Robertson DA, Savva GM, Kenny RA. Frailty and cognitive impairment–a review of the evidence and causal mechanisms. Ageing Res Rev. 2013;12(4):840–51. [DOI] [PubMed] [Google Scholar]
- 40.Kelaiditi E, Cesari M, Canevelli M, et al. Cognitive frailty: rational and definition from an (I.A.N.A./I.A.G.G.) international consensus group. J Nutr Health Aging. 2013;17(9):726–34. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Zhang Y, Xia H, Jiang X, Wang Q, Hou L. Prevalence and Outcomes of Cognitive Frailty Among Community-Dwelling Older Adults: A Systematic Review and Meta-Analysis. Res Gerontol Nurs. 2024;17(4):202–12. [DOI] [PubMed] [Google Scholar]
- 42.Dong P, Cheng C, Yin W, et al. Frailty as a mediator between sleep quality and cognitive impairment among the rural older adults: a cross-sectional study. BMC Geriatr. 2025;25(1):7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Kaur S, Banerjee N, Miranda M, et al. Sleep quality mediates the relationship between frailty and cognitive dysfunction in non-demented middle aged to older adults. Int Psychogeriatr. 2019;31(6):779–88. [DOI] [PubMed] [Google Scholar]
- 44.Li SB, Damonte VM, Chen C, et al. Hyperexcitable arousal circuits drive sleep instability during aging. Science. 2022;375(6583):eabh3021. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.Jia X, Gao W, Hagelin H, et al. Plasma proteomic signature of frailty in 50,506 adults. Cell Metab. 2026;38(5):1012–e10286. [DOI] [PubMed] [Google Scholar]
- 46.Wear D, Morrone CD, Simpson D, Liu F, Hussaini SA, Yu WH. Sex-specific acceleration of Alzheimer’s pathogenesis by chronic sleep-deprivation. Alzheimers Dement. 2026;22(2):e71099. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47.Hauglund NL, Andersen M, Tokarska K, et al. Norepinephrine-mediated slow vasomotion drives glymphatic clearance during sleep. Cell. 2025;188(3):606–e62217. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 48.Pu Y, Qi X, Huang L, et al. Gut metagenome and plasma metabolome profiles in older adults suggest pyruvate metabolism as a link between sleep quality and frailty. Cell Rep. 2025;44(10):116297. [DOI] [PubMed] [Google Scholar]
- 49.Piber D, Cho JH, Lee O, Lamkin DM, Olmstead R, Irwin MR. Sleep disturbance and activation of cellular and transcriptional mechanisms of inflammation in older adults. Brain Behav Immun. 2022;106:67–75. [DOI] [PubMed] [Google Scholar]
- 50.Pan Y, Ma L. Inflammatory Markers Associated with Physical Frailty and Cognitive Impairment. Aging Dis. 2024;16(2):859–75. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 51.Wennberg AMV, St Louis EK. Interconnectedness among frailty, sleep, and cognition: recent findings and clinical implications. Int Psychogeriatr. 2019;31(6):763–6. [DOI] [PubMed] [Google Scholar]
- 52.Piovezan RD, Abucham J, Dos Santos RV, Mello MT, Tufik S, Poyares D. The impact of sleep on age-related sarcopenia: Possible connections and clinical implications. Ageing Res Rev. 2015;23(Pt B):210–20. [DOI] [PubMed] [Google Scholar]
- 53.Qaseem A, Kansagara D, Forciea MA, Cooke M, Denberg TD, Clinical Guidelines Committee of the American College of Physicians. Management of Chronic Insomnia Disorder in Adults: A Clinical Practice Guideline From the American College of Physicians. Ann Intern Med. 2016;165(2):125–33. [DOI] [PubMed] [Google Scholar]
- 54.Corral-Pérez J, Vázquez-Sánchez MÁ, Casals-Sánchez JL, Contreras-García FJ, Costilla M, Casals C. A 6-month educational program improves sleep behaviour in community-dwelling frail older adults: A randomised controlled trial. Sleep Med. 2024;121:196–202. [DOI] [PubMed] [Google Scholar]
- 55.Dent E, Morley JE, Cruz-Jentoft AJ, et al. Physical Frailty: ICFSR International Clinical Practice Guidelines for Identification and Management. J Nutr Health Aging. 2019;23(9):771–87. [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
Data Availability Statement
The data underlying this study are available from the China Health and Retirement Longitudinal Study (CHARLS) database. Researchers may obtain access by registering and applying through the official CHARLS website at http://charls.pku.edu.cn.