Sleep duration and depressive symptoms among older Chinese adults: a serial mediation model of self-rated health and frailty Authors: Lingli Yang, Xiao Zheng, Run Wang, Ziying Fan, Benli Xue, Shujuan Xiao, Ke Tang, Mengjie Zhang, Yanming Liao, Chichen Zhang, Chunlan Zhou Published: 2026 ABSTRACT AbstractBackgroundDepressive symptoms in older adults compromise physical and mental health, burdening individuals, families, and society. Sleep duration, self-rated health (SRH), and frailty are critical correlates. Sleep shapes older adults’ health evaluation, and poor SRH elevates frailty risk. However, the interconnections among these four variables are rarely explored. Therefore, this study aimed to examine the serial mediating effect of SRH and frailty in the relationship between sleep duration and depressive symptoms in older adults.MethodsWe performed a cross-sectional study utilizing nationally representative data from the 2018 wave of the China Health and Retirement Longitudinal Study (CHARLS), including 5,111 participants aged ≥ 60 years. The SPSS PROCESS macro (Model 6) and Bootstrap were used to examine the serial mediating effect, with subgroup analyses stratified by sleep duration (≤ 6, 6–7, 7–8, > 8 h).ResultsSleep duration had a direct effect on depressive symptoms [B= -0.548, 95% CI = (-0.628, -0.468)], which accounted for 63.3% of the total effect. Sleep duration was also indirectly associated depressive symptoms through the following three pathways: (1) through SRH [B = -0.123, 95% CI = (-0.148, -0.098)], which accounted for 14.2% of the total effect; (2) through frailty [B = -0.108, 95% CI = (-0.137, -0.080)], which accounted for 12.5%; (3) through SRH and frailty [B = -0.088, 95% CI = (-0.106, -0.072)], which accounted for 10.2%. The total mediating effect was 36.7%. Subgroup analysis results indicated that the serial mediating effect was only present in the group with sleep duration ≤ 6 h.ConclusionsSelf-rated health and frailty have serial mediating effect between sleep duration and depressive symptoms in older adults. Shorter sleep duration correlates with poorer SRH, elevated frailty risk, and increased depressive symptoms. Healthcare professionals should monitor older adults with insufficient sleep and poor SRH to address potential frailty and depressive symptoms.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12877-026-07599-0.Keywords: Sleep duration, Depressive symptoms, Self-rated health, Frailty, Health management Introduction China is confronting a major social challenge of population aging. The proportion of individuals aged 60 and above is projected to exceed 30% by 2035, which will mark China’s entry into a deeply aging society [1]. Against this backdrop, many older adults in China are at risk of mental health problems, making this an important issue to be addressed in advancing healthy aging [2]. Depressive symptoms, a key indicator of mental well-being, are highly prevalent among older Chinese adults, with rates ranging from 20% to 60% [3, 4]. These symptoms are characterized by persistent low mood, anhedonia, pessimism, and hopelessness [5, 6]. Depressive symptoms not only severely reduce older adults’ quality of life [7] but also induce physical dysfunction [8] and may lead to suicide [4], while imposing substantial economic and care burdens on families and society [9]. Given the high prevalence and severe consequences of depressive symptoms, there is an urgent need to identify modifiable risk factors for depressive symptoms and elucidate the underlying mechanisms. Sleep duration is a well-established modifiable factor associated with depressive symptoms in older adults [10, 11]. Older adults who sleep less than six hours per night have a higher risk of depressive symptoms [12], and reduced sleep duration is associated with subsequent worsening of depressive symptoms [13]. However, most previous research has focused on direct physiological and biochemical mechanisms through which sleep duration affects depressive symptoms. Mental health is also influenced by both subjective and objective health status [14], but few existing studies have paid attention to the indirect role of these two factors (subjective and objective health) in the relationship between sleep duration and depressive symptoms. This neglect limits a comprehensive understanding of the relationship between sleep duration and depressive symptoms in later life. Therefore, it is necessary to explore how subjective and objective health factors mediate this association. Self-Rated Health (SRH), an individual’s subjective evaluation of overall health [15], may serve as the first mediator. Shorter sleep duration is associated with poorer SRH [16, 17], and older adults with subjective sleep difficulties tend to hold negative self-perceptions of aging [18]. Moreover, the association between SRH and depressive symptoms strengthens with age [19, 20], and poor SRH may precede and predict depressive symptoms in older adults [21]. Thus, SRH may be a mediator linking sleep duration to depressive symptoms, but this mediating role remains unexplored, thereby representing one of the research gaps that this study aims to address. Furthermore, according to the unified conceptual model of SRH proposed by Jylhä (2009), SRH is produced in a cognitive process that is inherently subjective and contextual, while also grounded in the biological and physiological state of the individual organism [22]. This model highlights that individuals can detect subclinical physiological dysregulation and latent health impairments via interoception, and integrate these subtle bodily signals into their global SRH, thereby reporting poorer SRH, which can predict subsequent adverse health outcomes [22]. Consequently, frailty, as an indicator of objective health status, may serve as the second mediator. Frailty is characterized by the progressive accumulation of multidimensional health deficits with aging, along with reduced physiological reserve and impaired stress resilience [23]. Such deficits become clinically recognizable or laboratory-diagnosable health impairments only when their accumulation reaches a certain threshold [23]. The more health deficits an individual accumulates, the higher the risk of developing frailty and the more severe the frailty status will be [23]. Interoception, as emphasized in the unified conceptual model of SRH [22], is intrinsically linked to this deficit accumulation process. Individuals capture the burden of sub-clinical conditions to report poor SRH [22, 24], which may predict poor objective health status—including frailty, as supported by longitudinal evidence [25]. Moreover, frailty is associated with depressive symptoms in older adults [26, 27] and also contributes to the persistence of depressive symptoms [28]. While insufficient sleep contributes to frailty and amplifies its adverse effects on depressive symptoms [29, 30]. Based on the above theoretical and empirical evidence, frailty may independently mediate the association between sleep duration and depressive symptoms, and SRH and frailty may sequentially form a serial mediating pathway. However, few studies have examined the mediating role of frailty in the association between sleep duration and depressive symptoms. More importantly, no studies have integrated subjective health assessment (SRH) and objective physical health (frailty) into a serial mediation model among older Chinese adults to clarify the pathway from sleep duration to depressive symptoms. These constitute two additional research gaps that this study intends to explore. Therefore, based on the aforementioned identified research gaps, we propose the following hypotheses: (1) Sleep duration has a direct effect on depressive symptoms; (2) Self-rated health has an indirect effect between sleep duration and depressive symptoms; (3) Frailty has an indirect effect between sleep duration and depressive symptoms; and (4) Self-rated health and frailty have a serial mediating effect in the relationship between sleep duration and depressive symptoms. Furthermore, some studies have indicated a non-linear association between sleep duration and depressive symptoms [31]. The official sleep health guidelines released by the National Office of the Patriotic Health Campaign recommend that older Chinese adults maintain six to seven hours of sleep per night [32]. And existing research among older Chinese adults has consistently shown that sleep duration of less than six hours is associated with a higher risk of depressive symptoms [33, 34]; However, the evidence regarding sleep duration exceeding eight hours remains inconclusive: some studies have reported a lower risk of depressive symptoms in this group [35], whereas others found no significant association [36]. Accordingly, this study further classified sleep duration into four groups (≤ 6, 6–7, 7–8, > 8 h) to verify whether the aforementioned hypotheses remain valid. Methods Samples and participants The data used in this study were obtained from the China Health and Retirement Longitudinal Study (CHARLS) [37]. The CHARLS conducted its baseline surveys between 2011 and 2012, which included individuals aged 45 and above and their family members, were carried out in 150 counties and 450 communities (including villages) across 28 provinces, autonomous regions, and municipalities directly under the Central Government, and covered 17,708 individuals, who resided in 10,257 households within 450 villages or urban communities [38, 39]. A biennial follow-up survey design is adopted by CHARLS, which has officially released data from five rounds of follow-up surveys. The survey years of these rounds are 2013, 2015, 2018, and 2020 respectively [39]. According to our study objectives and required variables, we used 2018 follow-up survey data, which collected 19,816 individual questionnaires from 11,628 families aged 45 and above [40]. This study included older adults aged 60 years and above, with participants excluded if they had missing values or logical errors for the dependent variable, independent variables, or covariates. Ultimately, this study included 5,111 effective samples after sample size screening. The procedure for screening the samples was as follows: (1) 8,762 participants under 60 years old were excluded; (2) 2,660 participants with missing values in depression scale were excluded; (3) 57 participants with missing values in the Frailty Index (FI) and SRH were excluded; (4) 25 participants with 0 sleep duration or logical errors were excluded; and 4) A total of 3,201 participants were excluded due to missing values or logical errors in covariates (including residence, empty nest, and financial support from children). The flow chart of sample selection is shown in Fig. 1. Flow chart of sample selection Measures Sleep duration Sleep duration, defined exclusively as nighttime sleep duration, was assessed using a single subjective self-report question from the CHARLS 2018 questionnaire: During the past month, how many hours of actual sleep did you get at night (average hours for one night)? This question is often used to assess an individual’s nighttime sleep duration [13, 41]. Self-rated health (SRH) A subjective question was used to evaluate SRH in the CHARLS 2018 questionnaire: Would you say your health is “very poor”, “poor”, “fair”, “good”, or “very good”? With answers ranging from “very poor” to “very good”, the scores ranged from 1 to 5, respectively. Moreover, a higher score indicates better SRH. Many studies use this single question to measure individuals’ SRH [20, 25, 42]. Frailty Frailty Index (FI) was used to evaluate frailty. FI, a commonly used health quantitative index in research and clinical practice, is based on the defect accumulation method, which combines multidimensional health variables into a continuous score that represents an individual’s overall personal health [23]. According to the construction principle of FI [43], variables must be related to health, the prevalence of health defects generally increases with age, health defects should not be prematurely saturated, and the selection of variables should be considered from a holistic health perspective. FI can be constructed using readily available information from health surveys [43]. Based on the CHARLS2018 questionnaire and referred to previously published literature [26, 44], thirty indicators of health-related deficits were selected to construct FI, including (1) 13 chronic diseases (hypertension, dyslipidemia, diabetes or high blood sugar, cancer or malignant tumor, chronic lung diseases, liver disease, heart attack, stroke, kidney disease, stomach or other digestive diseases, emotional, nervous, or psychiatric problems, arthritis or rheumatism, and asthma); (2) 5 types of disability (physical disabilities, brain damage/intellectual disability, vision problems, hearing problems, and speech impediment); and (3) 12 limitations related to activities of daily living (ADLs) and instrumental activities of daily living (IADLs). Details of the 30 variables in Supplementary Table 1. The options for 13 chronic diseases and 5 types of disability were “yes” or “no”, corresponding to scores of 1 or 0, and the options for 12 limitations of ADLs and IADLs were “I don’t have any difficulty”, “I have difficulty but can still do it or need help” or “I cannot do it”, corresponding to scores of 0, 0.5 or 1. The FI is calculated by dividing the total number of deficits by the 30 items, and then multiplying by 100 [45, 46]. A score of FI ranging from 0 to 100, and a higher FI indicates a greater degree of frailty [45, 46]. Depressive symptoms The 10-item version of the Centre for Epidemiological Studies Depression Scale (CESD-10) was used to assess depressive symptoms. This scale has 10 questions, and each question has 4 options: “Rarely or not”, “1 ~ 2 days per week”, “3 ~ 4 days per week”, and “5 ~ 7 days per week”. Each option has a score ranging from 0 to 3, with a total score ranging from 0 to 30. Questions 5 and 8 were reverse-scored prior to calculating the total score. A higher score indicates more severe depressive symptoms. The Cronbach’s α of the CESD-10 in this study was 0.810. Covariates The selection of covariates was based on literature reviews, including age (in years), sex (man, woman), education (illiteracy, primary school, middle school, high school and above), residence (city, combination zone between urban and rural areas, village), marital status (married, separation/divorced/widowed/never married), empty nest (yes, no), participation in social activities (yes, no), and financial support from children (yes, no). Statistical analysis Statistical analysis was conducted by IBM SPSS 26.0 software and SPSS PROCESS macro. The Kolmogorov‒Smirnov test was used to test the normality of continuous variables. The results showed that all continuous variables were not normally distributed, so medians and interquartile range (IQR) were used to describe continuous variables. Categorical variables are represented as percentages. A non-parametric test was used to compare depressive symptoms among older adults with different characteristics. Spearman’s rank correlation was used to determine the intercorrelations among sleep duration, SRH, frailty, and depressive symptoms. Harman’s single-factor test was used to test for common method bias for all four variables. No significant common method bias was observed in the data used in this study. We also conducted a multicollinearity test, and no multicollinearity was found. The serial mediating effect of SRH and frailty in sleep duration and depressive symptoms was examined by SPSS PROCESS macro software (model 6). In a serial mediation model, the “indirect effect” is essentially the product of regression coefficients across multiple paths, being non-normal distribution [47]. Bootstrap method does not require assuming that the data conform to a normal distribution, it can construct the empirical distribution of target statistics (e.g., the effect of each path in the serial mediation model) based on a large number of resampled pseudo-samples [47]. Therefore, we used 5000 bootstrap sampling to assess the mediating effects of the model and set a 95% bootstrap confidence interval (95% CI). Paths and mediating effects were considered significant when P < 0.05 and the 95% CI did not include zero. Moreover, to verify the stability of the mediation model, we conducted tests using three methods: adjusting the number of Bootstrap samples from 5,000 to 10,000; constructing two simple mediation models for comparative verification; and multiple imputation was employed to handle missing covariate data, and the serial mediation model was re-analyzed (N = 8312). We further categorized sleep duration into four groups: ≤6 h, 6–7 h (inclusive of 7 h), 7–8 h (inclusive of 8 h), and > 8 h. We then conducted four serial mediation models [≤6 h: Model 2; 6–7 h (inclusive of 7 h): Model 3; 7–8 h (inclusive of 8 h): Model 4; >8 h: Model 5] to examine the differences in mediating effect across different sleep duration intervals. Ethical considerations The CHARLS obtained ethical approval from the Institutional Review Board at Peking University (Ethics number: IRB00001052-11015). Our research team applied for authorization and downloaded the data. Results General characteristics of study participants Among the 5,111 older adults (median age was 68 years), 52.6% were men, and 47.4% were women. 47.3% of the older adults had received primary education, 73.5% lived in the village, 71.6% were married, 76.4% were empty nesters, 51.4% participated in social activities, and 84.5% received financial support from their children. The median sleep duration, SRH, frailty status, and depressive symptoms of the participants were 6.00 h, 3.00 points, 15.00 points, and 7.00 points, respectively. The details are shown in Table 1. Table 1General characteristics of study participantsVariablesN (%)/Medians (IQR)Age, years68 (64, 73)Sex Man2688 (52.6%) Women2423 (47.4%)Education Illiteracy1224 (23.9%) Primary school2415 (47.3%) Middle school891 (17.4%) High school and above581 (11.4%)Residence City1025 (20.1%) Combination zone between urban and rural areas328 (6.4%) Village3758 (73.5%)Marital status Married3657 (71.6%) Separation/Divorced/Widowed/Never married1454 (28.4%)Empty nest Yes3903 (76.4%) No1208 (23.6%)Participate in social activities Yes2629 (51.4%) No2482 (48.6%)Financial support from children Yes4320 (84.5%) No791 (15.5%)Sleep duration (hours)6.00 (5.00, 8.00)SRH3.00 (2.00, 3.00)Frailty15.00 (8.33, 25.00)Depressive symptoms7.00 (3.00, 13.00)IQR Interquartile Range, SRH Self-Rated Health, FI is calculated as (total deficits / 30 items) × 100; The score ranges of the CESD-10, SRH, and FI are 0–30, 1–5, and 0–100, respectively General characteristics of study participants IQR Interquartile Range, SRH Self-Rated Health, FI is calculated as (total deficits / 30 items) × 100; The score ranges of the CESD-10, SRH, and FI are 0–30, 1–5, and 0–100, respectively Nonparametric test and correlation analysis The results of the non-parametric test are shown in Table 2. Sex, education, residence, marital status, and participation in social activities were significantly different between groups of older adults with different levels of depression symptoms (P < 0.001). Four variables (sleep duration, SRH, frailty, and depressive symptoms) were significantly intercorrelated (P < 0.001). Specifically, sleep duration was positively associated with SRH, and frailty was positively associated with depressive symptoms; meanwhile, both sleep duration and SRH were negatively associated with frailty and depressive symptoms, respectively. As shown in Table 3. Table 2Nonparametric test of depressive symptoms and participants’ characteristicsVariablesN (%)Depressive symptoms P Median(P50)IQR(P25 - P75)Age, years68 (64, 73)6864–73-Sex< 0.001 Man2688 (52.6%)6.003.00–12.00 Women2423 (47.4%)9.004.00–15.00Education< 0.001 Illiteracy1224 (23.9%)10.005.00-16.75 Primary school2415 (47.3%)8.004.00–14.00 Middle school891 (17.4%)6.003.00–11.00 High school and above581 (11.4%)5.002.00–9.00Residence< 0.001 City1025 (20.1%)6.003.00–10.00 Combination zone between urban and rural areas328 (6.4%)6.002.00–11.00 Village3758 (73.5%)8.004.00–14.00Marital status< 0.001 Married3657 (71.6%)7.003.00–13.00 Separation/Divorced/Widowed/Never married1454 (28.4%)9.004.00–15.00Empty nest0.561 Yes3903 (76.4%)8.003.00–14.00 No1208 (23.6%)7.003.00–13.00Participate in social activities< 0.001 Yes2629 (51.4%)7.003.00–12.00 No2482 (48.6%)8.004.00–15.00Financial support from children0.150 Yes4320 (84.5%)8.003.00–13.00 No791 (15.5%)7.003.00–13.00IQR: Interquartile Range Nonparametric test of depressive symptoms and participants’ characteristics IQR: Interquartile Range Table 3Spearman’s rank correlation among four variablesVariablesSleep durationSRHFrailtyDepressive symptomsSleep duration-SRH0.196***-Frailty-0.223***-0.548***-Depressive symptoms-0.277***-0.414***0.435***-***P < 0.001; SRH Self-Rated Health, FI is calculated as (total deficits / 30 items) × 100; The score ranges of the CESD-10, SRH, and FI are 0–30, 1–5, and 0–100, respectively; A higher score of SRH indicated better SRH, while higher scores of the FI reflected greater frailty severity Spearman’s rank correlation among four variables ***P < 0.001; SRH Self-Rated Health, FI is calculated as (total deficits / 30 items) × 100; The score ranges of the CESD-10, SRH, and FI are 0–30, 1–5, and 0–100, respectively; A higher score of SRH indicated better SRH, while higher scores of the FI reflected greater frailty severity Serial mediation analyses Process Model 6 was used for the data analysis. A higher score of SRH indicated better SRH, while higher scores of the FI and CESD-10 reflected more severe frailty and more severe depressive symptoms. In the model, sleep duration was the independent variable, depressive symptoms were the dependent variable, SRH and frailty were the serial mediating variables, and gender, age, education, residence, marital status, empty nest status, participation in social activities, and financial support from children were covariates. The results showed that sleep duration had a significant direct effect on depressive symptoms (B=-0.548, P < 0.001), was significantly positively associated with SRH (B = 0.094, P < 0.001), and was negatively associated with frailty (B=-0.668, P < 0.001). This indicated that each 1-hour decrease in sleep duration was associated with a 0.548-point increase in depressive symptoms scores (range: 0–30), a 0.094-point decrease in SRH scores (range: 1–5), and a 0.668-point increase in frailty index scores (range: 0-100). And SRH was significantly negatively associated with frailty and depressive symptoms (B=-5.800; B=-1.309, P < 0.001); Frailty was also significantly positively associated with depressive symptoms (B = 0.161, P < 0.001). This indicated that each 1-point decrease in SRH scores was associated with a 5.800-point increase in frailty index scores and a 1.309-point increase in depressive symptoms scores, and each 1-point increase in frailty index scores was associated with a 0.161-point increase in depressive symptoms scores. Table 4 shows the results of the regression analysis of the serial mediation model. The path structure of the model is visually presented in Fig. 2. Table 4Results of regression analysis in serial mediation modelCriterionPredictors R R 2 FBt95%CISRHSleep duration0.2250.05130.2100.094***13.402(0.080, 0.108)Depressive symptomsSleep duration0.3880.150100.261-0.866***-19.536(-0.935, -0.779)FrailtySleep duration0.5770.333254.700-0.668***-9.219(-0.810,-0.526)SRH----5.800***-40.742(-6.079, -5.521)Depressive symptomsSleep duration0.5600.314212.062-0.548***-13.405(-0.628, -0.468)SRH----1.309***-14.273(-1.488, -1.129)Frailty---0.161***20.558(0.146, 0.176)*** P < 0.001; CI Confidence interval, B Unstandardised coefficient, SRH Self-Rated Health, Sleep duration: The unit of measurement is hours; FI is calculated as (total deficits / 30 items) × 100; The score ranges of the CESD-10, SRH, and FI are 0–30, 1–5, and 0–100, respectively; A higher score of SRH indicated better SRH, while higher scores of the FI reflected greater frailty severity Results of regression analysis in serial mediation model *** P < 0.001; CI Confidence interval, B Unstandardised coefficient, SRH Self-Rated Health, Sleep duration: The unit of measurement is hours; FI is calculated as (total deficits / 30 items) × 100; The score ranges of the CESD-10, SRH, and FI are 0–30, 1–5, and 0–100, respectively; A higher score of SRH indicated better SRH, while higher scores of the FI reflected greater frailty severity The serial mediating effects of self-rated health and frailty on sleep duration and depressive symptoms. *** P<0.001 To test the effect of each path in the serial mediation model, we employed the Bootstrap method with 5,000 repeated resampling. Results revealed that the 95% CI of each path excluded zero, demonstrating all paths were statistically significant (see details in Table 5). Regarding the effect decomposition of sleep duration on depressive symptoms: the total effect, direct effect, and indirect effect (mediated by SRH and frailty) all showed significant negative trends, with the total indirect effect accounting for 36.7% of the total effect. This indicates that insufficient sleep duration is not only directly associated with a higher risk of depressive symptoms, but also further elevates this risk through two indirect pathways: poor SRH and more severe frailty. Among the three indirect paths, all exhibited negative effects, with the path “sleep duration → SRH → depressive symptoms” contributing the most to the indirect effect, followed by “sleep duration → frailty → depressive symptoms” and “sleep duration → SRH → frailty → depressive symptoms” (specific percentage values are provided in Table 5). The robustness of the serial mediation model was tested, and the results of the serial mediation model were found to be reliable. The detailed results are presented in the supplementary material (supplementary Tables 2–5). Table 5The effects of sleep duration on depressive symptoms with SRH and frailty as mediatorsModel pathwaysBSEMediating effect95%CITotal effect sleep duration → depressive symptoms-0.866a0.044100%(-0.953, -0.779)Direct effect sleep duration → depressive symptoms-0.548a0.04163.3%(-0.628, -0.468)Total indirect effect sleep duration → depressive symptoms-0.318a0.02436.7%(-0.367, -0.272)Path 1: Sleep duration → SRH → depressive symptoms-0.123a0.01314.2%(-0.148, -0.098)Path 2: Sleep duration → frailty → depressive symptoms-0.108a0.01412.5%(-0.137, -0.080)Path 3: Sleep duration → SRH → frailty → depressive symptoms-0.088a0.00910.2%(-0.106, -0.072)a The 95% CI does not include 0; CI Confidence interval, B Unstandardised coefficient, SE Standard error, SRH Self-Rated Health, Sleep duration: The unit of measurement is hours; FI is calculated as (total deficits / 30 items) × 100; The score ranges of the CESD-10, SRH, and FI are 0–30, 1–5, and 0–100, respectively*** P < 0.001 The effects of sleep duration on depressive symptoms with SRH and frailty as mediators a The 95% CI does not include 0; CI Confidence interval, B Unstandardised coefficient, SE Standard error, SRH Self-Rated Health, Sleep duration: The unit of measurement is hours; FI is calculated as (total deficits / 30 items) × 100; The score ranges of the CESD-10, SRH, and FI are 0–30, 1–5, and 0–100, respectively *** P < 0.001 Additionally, we further analyzed whether the serial mediation effect still holds across different sleep duration intervals. The results showed that the serial mediation effect of SRH and frailty was only found in the group with sleep duration ≤ 6 h (Model 2), and the results of this group were basically consistent with those of the baseline model without stratification by sleep duration, indirectly verifying the stability of the model. In contrast, the serial mediation effect was not found in the groups with sleep duration of 6–7 h (including 7 h, Model 3), 7–8 h (including 8 h, Model 4), or > 8 h (Model 5). However, frailty still has a significant mediating effect between sleep duration and depressive symptoms in Model 5; meanwhile, frailty is positively associated with sleep duration. Detailed results are presented in the supplementary Tables 6–7. Discussion This study found that sleep duration not only had a direct effect on depressive symptoms but also had indirect effects through SRH, frailty, and the serial mediating effect of SRH and frailty, supporting all four hypotheses. Sleep duration was significantly negatively associated with depressive symptoms, supporting hypothesis (1), indicating that older adults with insufficient sleep at night was associated with a higher likelihood of depressive symptoms. Considering that previous studies have reported that both insufficient and excessive sleep are associated with depressive symptoms in older adults [31, 48], we further conducted subgroup analysis. The results showed that the serial mediation model was statistically significant only in the group with sleep duration ≤ 6 h, observed significance may be influenced by factors such as reduced statistical power. Notably, some previous findings also suggesting that depressive symptoms in older adults are primarily associated with insufficient sleep, which is consistent with our findings [12, 49]. The association between insufficient sleep and depressive symptoms may be related to the following physiological and functional factors. Neurotransmitter imbalance (e.g., inhibiting serotonin synthesis/release and increasing cortisol levels) and elevated inflammatory cytokines (CRP, IL-6) are associated with insufficient sleep [50]. Additionally, physical fatigue, mobility, and functional limitations in older adults are also linked to insufficient sleep [51]. Collectively, these factors are also associated with depressive symptoms [50, 51]. Additionally, insufficient sleep is also associated with depressive symptoms via subjective health assessments. We found that SRH played a mediating role between sleep duration and depressive symptoms, supporting hypothesis(2). SRH is an individual’s subjective assessment of self-health status based on the health information that is available to them [52]. Insufficient sleep is associated with subjective feelings such as fatigue, dizziness, and headache, and these unpleasant experiences are closely linked to poorer SRH among older adults. Age-related declines in the ability to maintain sleep quality are characterized by increased sleep fragmentation and reduced total sleep time [53]. Among Chinese older adults, ,the prevalence of insufficient sleep increased from 5.29% to 8.37%, whereas long sleep duration decreased from 28.77% to 19.27% between 2008 and 2018 [54]. In the culture of traditional Chinese medicine (TCM) health preservation, the principle of “regular daily routines” has profoundly shaped the health perceptions of Chinese older adults [55]. Therefore, older adults may report poor SRH when their sleep duration shortens. Conversely, for longer sleep duration, most Chinese older adults hold the perception that “being able to sleep well is a blessing”, viewing adequate rest and recuperation as essential for healthy aging, with prolonged rest serving as an important means of restoring energy [55]. Thus, the observation that the serial mediating effect was statistically significant only in the group with sleep duration ≤ 6 h may be also related to such culture‑based cognitive differences. However, the underlying physiological mechanisms warrant further investigation. This study identified frailty as a mediator between sleep duration and depressive symptoms, supporting Hypothesis(3). Sleep duration was negatively correlated with frailty severity, while frailty severity was positively correlated with depressive symptoms, findings that are consistent with previous research findings [26, 27]. Inflammatory processes are linked to the association between frailty and depressive symptoms [56]. Furthermore, frailty is closely associated with physical functional decline, multimorbidity, and chronic pain [57]. Such poor overall underlying health status may be further linked to depressive symptoms. However, in the group with sleep duration > 8 h, frailty also played a mediating role between sleep duration and depressive symptoms, and a positive association was observed between sleep duration and frailty. This suggests a non-linear relationship between sleep duration and frailty, which is consistent with the existing studies [58]. Longer sleep duration may be a marker of underlying health problems, e.g., chronic illness, low physical activity, or subclinical disease, and these factors are associated with frailty [59], further associated with depressive symptoms. Furthermore, against the cultural backdrop of “family-based elderly care” in China, older adults often worry that the decline in their functional capabilities will become a burden to their children—this may constitute a cultural factor underlying the association between frailty and depressive symptoms [57]. We identified serial mediating effects of SRH and frailty in the association between sleep duration and depressive symptoms, which supports Hypothesis(4). Notably, these serial mediating effects were only observed in the subgroup with sleep duration ≤ 6 h. The potential reason underlying the association between SRH and frailty may be related to the abnormal elevation of inflammatory levels [60–62]. Poor SRH associated with chronic insufficient sleep may reflect the deterioration of objective health status: it serves as a marker of subclinical states and reflects early changes that precede the accumulation of measurable physiological deficits, such as high blood pressure, early stages of diabetes, and mental stress [24, 25]. This indicates that the shorter the sleep duration, the poorer the SRH of older adults, which is associated with increased frailty and worsening depressive symptoms.Therefore, priority should be given to older adults with sleep duration ≤ 6 h. Although long sleep (> 8 h) is not associated with poor SRH, it is related to the frailty and thus linked to depressive symptoms, which also merits attention. However, as this study was a cross-sectional design, we cannot rule out the existence of dynamic interrelationships among these four variables, nor can we exclude the possibility of reverse causality. Depressive symptoms may lead to shortened or prolonged sleep duration in older adults through mechanisms such as neurotransmitter imbalance and sleep structure disruption, altered sleep duration may be a symptom of depression [63]. Meanwhile, depressive symptoms or frailty may also cause older adults to report poor SRH, and depressive symptoms may also worsen SRH and contribute to frailty. Future studies could use longitudinal data to further explore causal relationships between sleep duration, SRH, frailty, and depressive symptoms. Implications for public health Furthermore, this study found that older adults who are female, illiterate, reside in rural areas, have a marital status of separated/divorced/widowed/unmarried, or do not participate in social activities were more likely to have depressive symptoms. Therefore, elderly health services and depressive symptom screening should prioritize this population. Additionally, regarding the serial mediating effects of SRH and frailty between sleep duration and depressive symptoms, more targeted health management measures can be designed around this model in the future, especially for older adults with insufficient sleep. Notably, sleep duration may serve as a simple screening marker for depressive symptoms and vulnerability in older adults. Practical strategies could include conducting community-based sleep health programs (such as lectures, promotional brochures, and sleep interventions) to enhance older adults’ awareness of healthy sleep, helping them correctly evaluate their own sleep health, and reducing negative perceptions of SRH caused by sleep problems at the source; carrying out community frailty screening based on health records and popularizing knowledge about frailty, enabling older adults to master the “early signs of frailty” and “daily prevention methods”. Efforts should not be limited to “improving sleep” alone; instead, healthcare professionals should also enhance subjective health assessment (SRH) and implement objective physical health (frailty) interventions for older adults. This comprehensive intervention pathway is highly aligned with the core concept of healthy aging [64]. Limitations There are several limitations in this study. Firstly, this study adopts a cross-sectional design, which is unable to clarify the direction of causal associations. Future longitudinal studies can explore the causality and dynamic interrelationships among these four variables. Secondly, sleep duration and depressive symptoms were assessed subjectively via questionnaires rather than objectively measured by instruments, which may lead to recall biases and response bias. It is suggested that future studies could adopt a mixed measurement approach combining objective monitoring and subjective diaries to reduce biases through cross-validation between the two methods and improve data accuracy. Thirdly, given that all data in this study were sourced from Chinese older adults collected by the CHARLS, the interpretation of results should be cautious in other cultural contexts. Fourthly, although we controlled for several covariates in the mediation model, residual confounding may still exist (e.g., medication use, socioeconomic status, pain, or sleep quality). Future studies could further expand the range of covariates and incorporate the residual factors into the analysis. Finally, this study found that the serial mediation model was only valid in group with sleep duration ≤ 6 h, which might also be related to sleep duration categorization, more research is needed to explore this topic in the future. Conclusion This study revealed that insufficient sleep was associated with depressive symptoms in older adults directly, and correlated with depressive symptoms indirectly through SRH and frailty. Furthermore, subgroup analysis results indicated that this mediating effect was only present in the group with sleep duration ≤ 6 h. This suggests that healthcare professionals should identify insufficient sleep in older adults and carry out appropriate interventions, which is critical for the prevention of depressive symptoms in older adults, especially for those with poor SRH associated with insufficient sleep. By maintaining an optimal sleep duration, older adults may have good SRH and a reduced risk of frailty, with a corresponding reduction in the occurrence of depressive symptoms and further supporting healthy aging. Supplementary Information Supplementary Material 1. Acknowledgements The authors thank all older adults who participated in the study and staffs who carried out questionnaire surveys, and we also thank the China Health and Retirement Longitudinal Study (CHARLS) for sharing the data for us. Abbreviations Self-rated health China Health and Retirement Longitudinal Study Frailty Index Activities of daily living Instrumental activities of daily living The 10-item version of the Centre for Epidemiological Studies Depression Scale Confidence interval C-reactive protein Interleukin-6 Unstandardised coefficient Standard error Traditional Chinese Medicine Authors’ contributions Lingli Yang: Conceptualization; Data curation; Formal analysis; Visualization; Writing - original draft; Writing-review &editing; Xiao Zheng: Data curation; Validation; Writing-review & editing; Run Wang: Conceptualization; Writing - original draft; Writing-review & editing; Ziying Fan: Writing - original draft; Benli Xue: Formal analysis; Shujuan Xiao: Methodology; Ke Tang: Data curation; Mengjie Zhang: Writing - original draft; Yanming Liao: Visualization; Chichen Zhang: Conceptualization; Supervision; Funding acquisition; Writing - original draft; Writing-review & editing; Chunlan Zhou: Conceptualization; Supervision; Funding acquisition; Writing - original draft; Writing-review & editing. Funding This study supported the National Natural Science Foundation of China [Grant number: 72474093; 72274091]. Data availability The datasets used during the current study are available from the corresponding author on reasonable request (Chunlan Zhou, E-mail: lanchun200488@126.com; Chichen Zhang, E-mail: zhangchichen@sina.com). Declarations Ethics approval and consent to participate The China Health and Retirement Longitudinal Study (CHARLS) obtained ethical approval from the Institutional Review Board at Peking University (Ethics number: IRB00001052-11015). All participants volunteered to participate in the study and signed an informed consent form. 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. Lingli Yang, Xiao Zheng and Run Wang contributed equally to this work. Contributor Information Chichen Zhang, Email: zhangchichen@sina.com. Chunlan Zhou, Email: lanchun200488@126.com. References Associated Data Supplementary Materials Supplementary Material 1. Data Availability Statement The datasets used during the current study are available from the corresponding author on reasonable request (Chunlan Zhou, E-mail: lanchun200488@126.com; Chichen Zhang, E-mail: zhangchichen@sina.com).