Abstract
Background
Cognitive decline and depressive symptoms often co-occur in older adults, but research on the trajectories of these two conditions and their interrelationships remains limited.
Methods
Using data from 2,823 adults aged ≥ 60 years in the China Health and Retirement Longitudinal Study (CHARLS, 2011–2018), we identified depressive-symptom and cognitive-function trajectories using group-based trajectory modeling (GBTM) and then used a group-based dual-trajectory modeling (GBDTM) to quantify their interdependence by estimating conditional and joint probabilities across trajectory groups. Multinomial logistic regression was used to examine how baseline characteristics predicted the likelihood of membership in each trajectory group.
Results
Three distinct trajectories were identified for depressive symptoms (low: 31.0%, moderate: 46.2%, high: 22.8%) and cognitive function (high: 47.8%, declining: 34.6%, low: 17.6%). Depressive symptom trajectories mainly differed by symptom level with small fluctuations, whereas cognitive trajectories differed more in decline patterns, with modest decline in the high-functioning group and steeper declines in the other groups. Conditional and joint probabilities showed clear cross-domain linkage between depressive-symptom and cognitive-function trajectories. Dual-trajectory analysis revealed that 61.1% of participants in the low depressive symptoms group had high cognitive function, while 49.1% of the low cognition group had moderate depressive symptoms. The joint probability results showed that older adults with milder depressive symptoms were more likely to maintain high cognitive function. Higher education, better ADL function, and self-rated health were associated with favorable trajectory membership (P < 0.05).
Conclusion
Depressive symptoms and cognitive function were negatively associated among older Chinese adults. Interventions targeting depressive symptoms may help preserve cognitive health in aging populations.
Supplementary Information
The online version contains supplementary material available at 10.1186/s12889-026-26368-7.
Keywords: Depressive symptoms, Cognitive function, Older adults, Trajectory modelling
Introduction
With the intensification of population aging, cognitive decline and depressive symptoms among older adults have increasingly drawn attention and have become important public health issues affecting their quality of life. Depression contributes to emotional distress, family discord, and functional decline, worsening the prognosis of various illnesses and increasing mortality rates [1]. According to the World Health Organization (WHO), around 280 million people worldwide suffer from depression, including 5.7% of adults over 60 [2]. A report on the mental health of older adults in China reveals that 26.4% experience depressive symptoms to varying degrees, with 6.2% showing moderate to severe symptoms [3].
Cognitive function broadly refers to the mental processes used to acquire knowledge, process information, and reason. It encompasses domains such as perception, memory, learning, attention, decision making, and language [4]. Cognitive impairment is the decline in one or more cognitive functions and is classified as mild cognitive impairment (MCI) or dementia based on severity [5]. The global dementia population is expected to grow from 55 million in 2019 to 139 million by 2050. Moreover, dementia-related healthcare costs are projected to more than double by 2030, rising from US$1.3 trillion to US$2.8 trillion annually [6]. In China, the number of dementia cases among individuals aged 60 and above is estimated to reach around 48.98 million by 2050 [7].
The link between depressive symptoms and cognitive function has been a key focus in epidemiological research. Studies indicate that depressive symptoms correlate with a higher risk of cognitive impairment [8, 9]. On one hand, Persistent depressive symptoms are associated with a pattern of faster cognitive function score decline [10]. A history of depression may confer an increased risk for later developing dementia [8]. On the other hand, cognitive impairment may also raise the likelihood of developing depressive symptoms, as older adults with impaired cognitive function are more prone to feelings of helplessness, despair, and loneliness, which may lead to depressive symptoms. In older adults, depressive symptoms and cognitive decline often coexist, leading to emotional and physical challenges that affect daily functioning and reduce quality of life [11].
In addition to the interactive relationship between depressive symptoms and cognitive function, existing studies have also extensively explored the predictive factors of their changes over time. Previous studies have shown that sociodemographic characteristics (such as age, sex, and education level), health conditions, and behavioral factors (such as chronic status and social interaction) may be associated with the trajectories of depressive symptoms and cognitive function in older adults [12–15]. Identifying these factors helps delineate the baseline demographic, behavioral, and health characteristics associated with membership in different depressive-symptom or cognitive-function trajectories among older adults.
Previous studies have explored the longitudinal relationship between depressive symptoms and cognitive function. Aichele et al. [16] used a latent change model to analyze the temporal coupling effects between the two, but only focused on the average level of bidirectional association and did not reveal between individuals. Graziane et al. [17] used a dual-trajectory model to identify concurrent longitudinal trajectories of depressive symptoms and cognition function and revealed their impact on health outcomes, however they did not further analyze the factors associated with trajectory assignment or capture the interaction process between variables. Zhang et al. [18] constructed five trajectories of depressive symptoms based on a Chinese population and analyzed their impact on cognitive decline, but only modeled depressive trajectories and did not examine how depressive symptoms and cognitive function change together over time, and did not explore potential predictive factors.
Although previous studies examined depressive symptoms or cognitive function trajectories separately, we used GBTM to identify distinct trajectories of depressive symptoms and cognitive function separately, and then employed a group-based dual trajectory modeling (GBDTM) approach to identify the joint patterns of change of cognitive function and depressive symptoms in older adults. Group-Based Trajectory Modeling (GBTM) is a semi-parametric method used to identify trajectory groups within the analytic sample with similar change patterns in longitudinal data [19, 20]. GBTM uses longitudinal data to provide a model-based summary of heterogeneity in outcome trajectories within the analytic sample by grouping individuals with similar patterns over time [21]. This method is typically applied to examine the longitudinal course of a single outcome variable and to identify subgroups that exhibit distinct patterns over time. Building on this, the GBDTM extends the application scope of GBTM to simultaneously analyze the trajectory relationships between two correlated outcome variables. The modelling process of GBDTM typically involves two steps: first, conducting independent GBTM analyses on the two variables to identify their respective longitudinal trajectory types; it captures symptom trajectories over time, focusing on changes in depressive symptoms and cognitive function. Second, estimating the joint probability and conditional probability between the trajectories of the two variables to reveal their patterns of association during the longitudinal process. After identifying distinct trajectory groups through GBTM, we further explored how baseline characteristics were associated with different dual-trajectory patterns of depressive symptoms and cognitive function. We applied multinomial logistic regression, using the trajectory group as the outcome and baseline sociodemographic, health, and behavioral factors as predictors to examine their associations with different trajectory memberships. This also provides the theoretical basis and practical directions for early intervention in high-risk populations. Therefore, further analyses of the potential factors associated with the trajectories of depressive symptoms and cognitive function changes in older adults are of great significance for promoting mental and cognitive health in older adults and advancing precision public health interventions.
Objects and methods
Sources
This study utilized data from the China Health and Retirement Longitudinal Study (CHARLS) follow-up survey, which targets middle-aged and older adults (45 + years) in China [22]. The initial baseline survey, conducted between June 2011 and March 2012, covered 450 villages/communities, 150 counties/districts, and 28 provinces. These locations were randomly chosen through a multistage stratified sampling strategy, with follow-ups every two years starting in 2011. The CHARLS study was approved by the Medical Ethics Review Committee of Peking University (IRB00001052-11015), and all participants provided informed consent.
This study initially included 4,848 participants aged 60 and older at baseline. Since GBTM requires data from at least three time points per participant, we applied a stricter inclusion criterion by including only those with complete data across all four waves (2011, 2013, 2015, and 2018) for depressive symptoms and cognitive function. After excluding participants with incomplete data, 2,823 individuals remained for the final analysis. Participants with missing values in other variables were also excluded. The detailed selection process is shown in Supplementary Fig. 1. Wave-specific response rates for CHARLS (2011–2018), as reported in the official CHARLS documentation, appear in Supplementary Table 1.
Research variables
Depressive symptoms
Depressive symptoms were measured using the 10-item Short Version of the Center for Epidemiological Studies Depression Scale (CESD-10) [23]. The scale consists of ten questions, each with four response options, scored from 0 to 3. Total scores range from 0 to 30 [24], Higher scores indicate more severe depressive symptoms.
Measurement of cognitive function
Based on similar measures used in the American Health and Retirement Study (HRS). It consists of two components: episodic memory and executive function [25]. Episodic memory is assessed by asking participants to recall 10 unrelated Chinese words, with accuracy measured immediately and several minutes later. Following the method of McArdle et al., we constructed an episodic memory measurement index by averaging the immediate recall and delayed recall scores, with a range from 0 to 10 [26]. Executive function is assessed through identifying dates, months, seasons, and days of the week, completing five subtraction tasks (subtracting 7 from 100), and imitating a drawing, with a maximum score of 11 points. The scores from both components are summed (0 to 21 points) to evaluate overall cognitive function, with higher scores indicating better performance. Although the total score differs from the 31-point scale used in some literature, the task design remains consistent, assessing both immediate and delayed recall abilities, ensuring good comparability in terms of directionality and trends. Additionally, this scoring method has been widely adopted in cognitive studies using CHARLS data, demonstrating good practicality and reliability in practice [27–29].
Variables
In this study, potential predictors were selected based on previous literature and correlation analyses. Eighteen variables were chosen for analysis, encompassing demographics, health status and behaviors. These included age, education, marital status, gender, medical insurance, chronic status, falls, hearing status, toothlessness, life satisfaction, sleep duration, social interaction, smoking, drinking, self-rated health, activities of daily living (ADL), and self-rated memory. Activities of Daily Living (ADL) [30] include six basic activities: bathing, dressing, using the toilet, moving indoors, maintaining personal hygiene, and eating. Score of 6 indicates normal function [31]. Other indicators are shown in supplementary Table 2.
Statistical analysis
In this study, the GBTM was used to examine the trajectories of depressive symptoms and cognitive functioning. The optimal model was chosen based on diagnostic statistics, including the Bayesian Information Criterion (BIC), Akaike Information Criterion (AIC), and the average posterior probability (AvePP). The smaller the AIC and BIC, the better. The average posterior probability must be at least greater than 0.7, and the closer to 1, the better [32]. Additionally, the GBDTM was applied to extend the GBTM methodology by identifying the longitudinal relationships between variables and estimating the conditional and joint probabilities of the two related trajectories. The fitting process consisted of several steps: first, univariate GBTM modeling was applied separately to depressive symptoms and cognitive functioning trajectories, using maximum likelihood estimation to determine the optimal number of trajectory groups. Then, based on the identified optimal groups, the polynomial order of each trajectory was adjusted to optimize their shapes. Using the final univariate model parameters, the dual-trajectory model was estimated to explore the longitudinal association between depressive symptoms and cognitive functioning. The models were fitted using the traj plug-in for Stata 17, and other statistical analyses were conducted in R 4.4.0.
Subsequently, to examine associations between baseline predictors and trajectory categories, this study used trajectory categories as the dependent variable and adopted multinomial logistic regression to simultaneously incorporate variables such as sociodemographic characteristics, health status, and functional status into the model for analysis. By estimating the independent associations of each variable on trajectory category affiliation while controlling for other factors, the study reported their relative risk ratio (RRR), 95% confidence intervals (CI), and P-values, with a significance level of α = 0.05. To assess the potential bias caused by missing data, we conducted a comparative analysis of baseline characteristics between the final sample and the excluded sample.
Results
Baseline characteristics of the study population
The baseline characteristics of the study population are shown in Table 1. The study included 2,823 participants aged 60 and older, with a mean age of 66.5 ± 5.4 years; 56.4% were female. The average depressive symptoms score for the study population was 10.0 ± 6.49, while the cognitive function score averaged 9.33 ± 4.20. Moreover, 36% of participants were illiterate, and 79.7% had chronic status. The proportion of women with depressive symptoms and reduced cognitive function is higher than that of men. The included and excluded samples showed significant differences in some baseline characteristics (see Appendix 1 Table 1), suggesting that data missingness may introduce systematic bias. We addressed potential bias from missing data by using multiple imputation by chained equations (MICE, m = 20 imputations) to impute missing covariate values and rerunning the analyses across the imputed datasets as a sensitivity check. The imputed results showed the same trajectory solutions and comparable associations to the primary analysis, which supports the robustness of our conclusions, although some selection bias may remain. We report the full results in Appendix 1.
Table 1.
Baseline characteristics of participants (mean 
standard deviation [
S], n [%])
| Variable | Population | Male | Female | P |
|---|---|---|---|---|
| sample size | 2823 | 1232 | 1591 | |
| age (years) | 66.5 ± 5.4 | 66.7 ± 5.3 | 66.4 ± 5.4 | 0.025 |
| hukou | 0.003 | |||
| agricultural | 2291(81.2) | 970(78.7) | 1322(83.1) | |
| non-agricultural | 531(18.8) | 262(21.3) | 269(16.9) | |
| education | < 0.001 | |||
| illiterate | 1015(36.0) | 210(17.0) | 805(50.6) | |
| primary and below | 1361(48.2) | 725(58.8) | 636(40.0) | |
| Junior high school and above | 447(15.8) | 297(24.1) | 150(9.4) | |
| marital status | < 0.001 | |||
| married living with spouse | 2229(79.0) | 1062(86.2) | 1167(73.4) | |
| others | 594(21.0) | 170(13.8) | 424(26.6) | |
| medical insurance | 0.658 | |||
| No | 2689(95.3) | 1176(95.5) | 1513(95.1) | |
| Yes | 134(4.7) | 56(4.5) | 78(4.9) | |
| chronic status | 0.066 | |||
| No | 574(20.3) | 270(21.9) | 304(19.1) | |
| Yes | 2249(79.7) | 962(78.1) | 1287(80.9) | |
| falls | 0.001 | |||
| No | 2234(79.1) | 1011(82.1) | 1223(76.9) | |
| Yes | 589(20.9) | 221(17.9) | 368(23.1) | |
| hearing state | 0.415 | |||
| good | 970(34.4) | 417(33.8) | 553(34.8) | |
| general | 1343(47.6) | 579(47.0) | 764(48.0) | |
| bad | 510(18.1) | 236(19.2) | 274(17.2) | |
| toothlessness | 0.522 | |||
| No | 2406(85.2) | 1056(85.7) | 1350(84.9) | |
| Yes | 417(14.8) | 176(14.3) | 241(15.1) | |
| life satisfaction | 0.521 | |||
| satisfactory | 2378(84.2) | 1032(83.8) | 1346(84.6) | |
| unsatisfactory | 444(15.7) | 200(16.2) | 244(15.3) | |
| sleep duration | 5.99 ± 2.03 | 6.17 ± 1.91 | 5.85 ± 2.10 | < 0.01 |
| social interaction | 0.891 | |||
| No | 1505(53.3) | 655(53.2) | 850(53.4) | |
| Yes | 1318(46.7) | 577(46.8) | 741(46.6) | |
| smoking | < 0.001 | |||
| No | 2020(71.6) | 551(44.7) | 1469(92.3) | |
| Yes | 803(28.4) | 681(55.3) | 122(7.7) | |
| drinking | < 0.001 | |||
| No | 2043(72.4) | 638(51.8) | 1405(88.3) | |
| Yes | 780(27.6) | 594(48.2) | 186(11.7) | |
| self-rated health | 0.008 | |||
| good | 450(15.9) | 222(18.0) | 228(14.3) | |
| general | 1458(51.6) | 640(51.9) | 818(51.4) | |
| bad | 915(32.4) | 370(30.0) | 545(34.3) | |
| ADL | 0.016 | |||
| No impaired | 2057(72.9) | 926(75.2) | 1131(71.1) | |
| impaired | 766(27.1) | 306(24.8) | 460(28.9) | |
| self-rated memory | 0.002 | |||
| good | 378(13.4) | 176(14.3) | 202(12.7) | |
| general | 1209(42.8) | 563(45.7) | 646(40.6) | |
| bad | 1236(43.8) | 493(40.0) | 743(46.7) | |
| depressive symptoms | 10.0 ± 6.49 | 9.01 ± 6.05 | 10.76 ± 6.72 | < 0.001 |
| cognitive function | 9.33 ± 4.20 | 10.55 ± 3.80 | 8.38 ± 4.25 | < 0.001 |
represents the sample mean; S represents the sample standard deviation
Data are from the CHARLS, 2011–2018. Total sample size = 2,823
Depressive symptoms trajectories and cognitive function trajectories
The number of trajectory groups was determined based on AIC/BIC values and model interpretability, while polynomial orders were selected according to p-values. We assessed model adequacy and classification quality using AIC, BIC, and average posterior probabilities (AvePP). The selected three-group solutions met our prespecified criterion for classification quality (AvePP > 0.7). Detailed fit statistics are reported in Supplementary Tables 3–4, and the corresponding diagnostics from the imputed sensitivity analyses are presented in Appendix 1 Tables 2 and 3. We modeled depressive symptoms and cognitive function trajectories separately. The fitted trajectories are shown in Supplementary Figs. 2–3, and the estimated parameters for each trajectory are reported in Supplementary Table 5.
Figure 1 illustrates the optimal groupings and trajectory patterns of depressive symptoms in older adults, divided into three groups based on depressive symptom trends: a low depressive symptom group (n = 875), a moderate depressive symptom group (n = 1304), and a high depressive symptom group (n = 643). The depressive-symptom trajectories were largely separated by overall symptom level, with only minor within-group fluctuations across cycles. Similarly, Fig. 2 depicts the optimal groupings and trajectory patterns of cognitive functioning, with three groups based on cognitive functioning trends: a low cognitive functioning group (n = 497), a cognitive decline group (n = 977), and a high cognitive functioning group (n = 1349). The cognitive function trajectories exhibited clearer differences in longitudinal change: cognitive scores declined in all groups over follow-up, but the magnitude of decline varied, with a modest decrease in the high-functioning group and steeper decreases in the cognitive-decline and low-functioning groups. The polynomial orders for the depressive symptom trajectories were 2, 3, and 3, while for the cognitive functioning trajectories, they were 1, 2, and 2.
Fig. 1.
Depressive symptoms trajectory. Note: Cycles 1, 2, 3 and 4 are for 2011-2012, 2013-2014, 2015-2017 and 2018-2019, respectively. trajT: trajectory of time. Data are from the CHARLS, 2011–2018. Total sample size = 2,823
Fig. 2.
Cognitive function trajectory. Note: Cycles 1, 2, 3 and 4 are for 2011-2012, 2013-2014, 2015-2017 and 2018-2019, respectively. trajT: trajectory of time. Data are from the CHARLS, 2011–2018. Total sample size = 2,823
Factors associated with trajectories of depressive symptoms and trajectories of cognitive function
The multinomial logistic regression results were derived from a multivariable model in which all potentially relevant covariates were included simultaneously to control for possible confounding. The reported relative risk ratios (RRRs), 95% confidence intervals, and p-values are all based on this fully adjusted model.
Factors associated with attribution of depressive symptoms trajectories
To describe how baseline characteristics differed across the depressive-symptom trajectory groups, we conducted multinomial logistic regression with trajectory membership as the categorical outcome. According to the multinomial logistic regression analysis (Table 2), several baseline characteristics were significantly associated with depressive symptom trajectories, using the high depressive symptom trajectory as the reference group. Interestingly, older adults tended to show lower levels of depressive symptoms (RRR = 1.031, 95% CI: 1.008–1.057) and moderate (RRR = 1.023, 95% CI: 1.002–1.043) depressive symptom trajectories. This pattern may reflect survivorship and psychological adaptation effects. Older individuals who remained in the cohort usually had stronger mental resilience and reported fewer emotional problems. Participants with an agricultural hukou were less likely to be in the low depressive group (RRR = 0.406, 95% CI: 0.286–0.575) compared with those with a non-agricultural hukou. Being married and living with a spouse (RRR = 1.404, 95% CI: 1.044–1.890) and being male (low group: RRR = 2.361, 95% CI: 1.726–3.230; moderate group: RRR = 1.606, 95% CI: 1.222–2.109) were associated with a higher likelihood of belonging to less severe depressive trajectories. Participants without chronic diseases (RRR = 1.449, 95% CI: 1.068–1.967) and those who had not experienced falls (RRR = 1.354, 95% CI: 1.009–1.816) were also more likely to belong to the low depressive group. Longer sleep duration was positively associated with being in the low (RRR = 1.316, 95% CI: 1.239–1.397) and moderate (RRR = 1.102, 95% CI: 1.049–1.157) groups. Participants with good or general self-rated health had markedly higher probabilities of belonging to less severe trajectories (low group: RRR = 7.678, 95% CI: 4.998–11.795; moderate group: RRR = 2.259, 95% CI: 1.527–3.342). Functionally independent individuals (ADL = No impaired) were also more likely to belong to the low (RRR = 3.082, 95% CI: 2.343–4.053) and moderate (RRR = 1.931, 95% CI: 1.561–2.387) depressive symptom trajectories compared with those with ADL impairment. Similarly, participants with good or general self-rated memory were more likely to belong to the low depressive symptom group (RRR = 2.113, 95% CI: 1.419–3.147; RRR = 2.233, 95% CI: 1.719–2.902).
Table 2.
Multinomial logistic regression analysis of depressive symptoms trajectory categorization
| Variable | Low depressive symptoms group RRR(95%CI) | Moderate depressive symptoms group RRR(95%CI) |
|---|---|---|
| age (years) | 1.031(1.008, 1.057)* | 1.023(1.002, 1.043)* |
| hukou | ||
| agricultural | 0.406(0.286, 0.575)* | 0.771(0.558, 1.068) |
| non-agricultural | Ref | Ref |
| education | ||
| illiterate | 0.762(0.498,1.165) | 0.877(0.597, 1.287) |
| primary and below | 0.700(0.478,1.025) | 0.897(0.630, 1.277) |
| junior high school and above | Ref | Ref |
| marital status | ||
| married living with spouse | 1.404(1.044,1.890)* | 1.109(0.868, 1.417) |
| others | Ref | Ref |
| gender | ||
| male | 2.361(1.726, 3.230)* | 1.606(1.222, 2.109)* |
| female | Ref | Ref |
| medical insurance | ||
| No | 0.781(0.443, 1.377) | 1.063(0.668, 1.690) |
| Yes | Ref | Ref |
| chronic status | ||
| No | 1.449(1.068, 1.967)* | 1.094(0.827, 1.447) |
| Yes | Ref | Ref |
| falls | ||
| No | 1.354(1.009, 1.816)* | 1.054(0.834, 1.332) |
| Yes | Ref | Ref |
| hearing state | ||
| good | 1.020(0.714,1.457) | 0.936(0.697, 1.258) |
| general | 1.044(0.750,1.453) | 1.044(0.803, 1.358) |
| bad | Ref | Ref |
| toothless | ||
| No | 1.246(0.894, 1.735) | 1.141(0.864, 1.506) |
| Yes | Ref | Ref |
| life satisfaction | ||
| satisfactory | 1.122(0.817,1.540) | 1.068(0.817, 1.395) |
| unsatisfactory | Ref | Ref |
| sleeptime | 1.316(1.239, 1.397)* | 1.102(1.049, 1.157)* |
| social interaction | ||
| No | 0.879(0.694, 1.113) | 0.883(0.721, 1.081) |
| Yes | Ref | Ref |
| smoking | ||
| No | 1.109(0.810,1.519) | 0.898(0.682, 1.182) |
| Yes | Ref | Ref |
| drinking | ||
| No | 1.209(0.900, 1.624) | 1.075(0.832, 1.389) |
| Yes | Ref | Ref |
| Self-assessment health | ||
| good | 7.678(4.998,11.795)* | 2.259(1.527, 3.342)* |
| general | 3.667(2.776,4.844)* | 1.776(1.430, 2.205)* |
| bad | Ref | Ref |
| ADL | ||
| No impaired | 3.082(2.343, 4.053)* | 1.931(1.561, 2.387)* |
| impaired | Ref | Ref |
| Self-assessment memory | ||
| good | 2.113(1.419,3.147)* | 1.291(0.894, 1.863)* |
| general | 2.233(1.719,2.902)* | 1.560(1.250, 1.947)* |
| bad | Ref | Ref |
*P < 0.05, Ref is the reference group. RRR Relative Risk Ratio. Data are from the CHARLS, 2011–2018. Total sample size = 2,823. Reference group: high depressive symptom group. RRR > 1 indicates a higher likelihood of belonging to the low or moderate depressive symptom trajectory relative to the high depressive symptom group
Factors associated with attribution of cognitive function trajectories
We further examined how baseline sociodemographic and health characteristics varied across the cognitive-function trajectory groups. According to the multinomial logistic regression analysis (Table 3), several baseline characteristics were significantly associated with cognitive function trajectories, with the high cognitive function trajectory serving as the reference category. Older participants were more likely to belong to the low cognitive function (RRR = 1.155, 95% CI: 1.124–1.188) and cognitive decline (RRR = 1.088, 95% CI: 1.065–1.112) trajectories. Individuals with an agricultural hukou were also more likely to be classified into the low (RRR = 5.044, 95% CI: 2.966–8.579) and decline (RRR = 3.015, 95% CI: 2.231–4.074) trajectories compared with those with a non-agricultural hukou. Educational attainment showed a clear gradient: illiterate participants had a markedly higher probability of belonging to the low (RRR = 29.197, 95% CI: 12.055–42.367) and decline (RRR = 22.430, 95% CI: 14.841–33.899) trajectories, and those with primary education or below also showed elevated risks (RRR = 10.298, 95% CI: 2.482–42.728; RRR = 3.945, 95% CI: 2.741–5.680) compared with those with junior high school or above. Being married and living with a spouse was negatively associated with low cognitive function (RRR = 0.707, 95% CI: 0.507–0.986). Male participants were less likely than females to belong to the low (RRR = 0.312, 95% CI: 0.211–0.462) and decline (RRR = 0.570, 95% CI: 0.441–0.738) trajectories. Participants without medical insurance were more likely to belong to the cognitive decline group (RRR = 1.714, 95% CI: 1.048–2.805). Participants reporting good hearing were less likely to experience cognitive decline (RRR = 0.707, 95% CI: 0.519–0.964). Individuals without regular social interaction were more likely to belong to the low (RRR = 1.981, 95% CI: 1.497–2.622) and decline (RRR = 1.505, 95% CI: 1.231–1.839) trajectories. Functionally independent participants (ADL = No impaired) were less likely to belong to the low (RRR = 0.692, 95% CI: 0.507–0.945) or declining (RRR = 0.692, 95% CI: 0.547–0.874) cognitive trajectories compared with those with ADL impairment. In addition, participants who rated their memory as good (RRR = 0.240, 95% CI: 0.147–0.393) or general (RRR = 0.310, 95% CI: 0.228–0.422) were less likely to belong to the low cognitive function trajectory, and similar patterns were observed for cognitive decline (good: RRR = 0.524, 95% CI: 0.375–0.732; general: RRR = 0.498, 95% CI: 0.399–0.621).
Table 3.
Multinomial logistic regression analysis of cognitive function trajectory categorization
| Variable | Low cognitive function group RRR (95% CI) | Cognitive decline group RRR (95%CI) |
|---|---|---|
| age (years) | 1.155(1.124, 1.188)* | 1.088(1.065, 1.112)* |
| hukou | ||
| agricultural | 5.044(2.966, 8.579)* | 3.015(2.231, 4.074)* |
| non-agricultural | Ref | Ref |
| education | ||
| illiterate | 29.197(12.055, 42.367)* | 22.430(14.841, 33.899)* |
| primary and below | 10.298(2.482, 42.728)* | 3.945(2.741, 5.680)* |
| junior high school and above | Ref | Ref |
| marital status | ||
| married living with spouse | 0.707(0.507, 0.986)* | 0.810(0.626, 1.047) |
| others | Ref | Ref |
| gender | ||
| male | 0.312(0.211, 0.462)* | 0.570(0.441, 0.738)* |
| female | Ref | Ref |
| medical insurance | ||
| No | 1.786(0.940, 3.393) | 1.714(1.048, 2.805)* |
| Yes | Ref | Ref |
| chronic status | ||
| No | 1.313(0.925, 1.865) | 1.265(0.979, 1.633) |
| Yes | Ref | Ref |
| falls | ||
| No | 1.121(0.802, 1.567) | 1.045(0.813,1.342) |
| Yes | Ref | Ref |
| hearing state | ||
| good | 0.666(0.442, 1.002) | 0.707(0.519,0.964)* |
| general | 0.745(0.513, 1.082) | 0.916(0.691,1.214)* |
| bad | Ref | Ref |
| toothless | ||
| No | 0.938(0.645, 1.363) | 1.093(0.821,1.455) |
| Yes | Ref | Ref |
| life satisfaction | ||
| satisfactory | 0.763(0.528, 1.104) | 0.781(0.592,1.030) |
| unsatisfactory | Ref | Ref |
| sleeptime | 0.973(0.911, 1.039) | 1.012(0.962, 1.064) |
| social interaction | ||
| No | 1.981(1.497, 2.622)* | 1.505(1.231,1.839)* |
| Yes | Ref | Ref |
| smoking | ||
| No | 0.881(0.596, 1.303) | 0.850(0.658,1.098) |
| Yes | Ref | Ref |
| drinking | ||
| No | 0.887(0.619, 1.271) | 0.875(0.686,1.117) |
| Yes | Ref | Ref |
| self-assessment health | ||
| good | 0.691(0.427, 1.116) | 1.113(0.796,1.556) |
| general | 0.884(0.644, 1.212) | 0.954(0.754,1.209) |
| bad | Ref | Ref |
| ADL | ||
| No impaired | 0.692(0.507, 0.945)* | 0.692(0.547,0.874)* |
| impaired | Ref | Ref |
| self-assessment memory | ||
| good | 0.240(0.147, 0.393)* | 0.524(0.375,0.732)* |
| general | 0.310(0.228, 0.422)* | 0.498(0.399,0.621)* |
| bad | Ref | Ref |
*P < 0.05, Ref is the reference group. RRR Relative Risk Ratio. Data are from the CHARLS, 2011–2018. Total sample size = 2,823. Reference group: high cognitive function group. RRR > 1 indicates a higher likelihood of belonging to the low cognitive function or cognitive decline trajectory relative to the high cognitive function group
A dual-trajectory model of depressive symptoms and cognitive function
The trajectories of depressive symptoms and cognitive function were analyzed using a dual-trajectory model. The conditional probabilities of their interrelationship are presented in Fig. 3, Fig. 3A presents the conditional probabilities of cognitive function trajectories given different depressive symptom trajectories. Fig. 3B presents the conditional probabilities of depressive symptom trajectories given different cognitive function trajectories. The joint probabilities are shown in Fig. 4. The conditional and joint probability results (Figs. 3 and 4) provide the main evidence on how depressive symptoms and cognitive function co-develop by linking trajectory-group membership across the two domains.
Fig. 3.
Conditional probabilities in a dual-trajectory model of depressive symptoms and cognitive function
Fig. 4.
Joint probability heat map in a dual-trajectory model of depressive symptoms and cognitive function
Figure 3A presents the conditional probabilities of cognitive function trajectories given different depressive symptom trajectories. In the low depressive symptoms group, individuals had the highest likelihood of maintaining high cognitive function (0.61) and the lowest probability of experiencing low cognitive function (0.11). In the moderate depressive symptoms group, the probability of high cognitive function was notable (0.45), though there was a more even distribution of probabilities across various cognitive states. In contrast, those with high depressive symptoms were most likely to experience cognitive decline, with a probability of 0.42.
Figure 3B presents the conditional probabilities of depressive symptom trajectories given different cognitive function trajectories. The highest probability of experiencing moderate depressive symptoms was found in the low cognitive functioning group (0.49), suggesting that individuals with lower cognitive functioning are most likely to experience these fluctuations. A similar trend was observed in the cognitive decline group (0.47), indicating substantial variations in depressive symptoms during cognitive decline. In contrast, the high cognitive functioning group had a notable probability of moderate depressive symptom (0.44), but the likelihood of high levels of depressive symptoms was the lowest (0.15) compared to the other two groups. This suggests that higher cognitive functioning may offer a protective effect, reducing the severity of depressive symptoms.
Figure 4 shows clear patterns in the joint distribution of depressive symptoms and cognitive function trajectories. The most frequent combination was moderate depressive symptoms with high cognitive function (joint probability = 0.21), followed by low depressive symptoms with high cognitive function (0.19). In contrast, few participants showed both low depressive symptoms and low cognitive function (0.04). These results suggest that individuals with better cognitive function tended to have fewer depressive symptoms, while those with low cognitive function often showed moderate to high levels of depressive symptoms.
Discussion
This study identified longitudinal patterns of depressive symptoms and cognitive functioning in older adults, highlighting the demographic, health, and behavioral factors linked to these trajectories. It also found a negative correlation between depressive symptoms and cognitive functioning, with the severity and stability of depressive symptoms being significantly associated with cognitive status. Moreover, the results describe how depressive symptoms and cognitive function co-occur over time, indicating a close temporal association between the two domains. The conditional and joint probabilities from the dual-trajectory model quantify the linkage between affective and cognitive trajectories by showing how membership in one set of trajectory groups maps onto membership in the other. This approach summarizes co-development patterns directly, rather than describing depressive-symptom and cognitive-function trajectories in parallel. These probability patterns provide a concise representation of cross-domain interdependence in longitudinal change.
Our study identified three trajectories of depressive symptom development in the Chinese elderly population: low depressive symptoms group、moderate depressive symptoms group、high depressive symptoms group. Murphy et al. [33] used the CESD-10 scale and analyzed four years of depressive symptoms scores, finding three trajectories: persistently non-depressed, mildly depressed with worsening, and persistently depressed with worsening. These findings are consistent with our results. Additionally, the study examined factors associated with depressive symptom trajectories. Age, gender, and marital status were found to be significant predictors of depressive symptoms, consistent with the findings of Noori et al. [34] Chronic status, ADLs, self-rated health, and sleep duration were also found to be associated with depressive symptoms, supporting Cui et al.’s findings [35]. Poor health increases the risk of falls among older adults [36], which may subsequently increase the likelihood of experiencing depressive symptoms. Older adults in agricultural households are more likely to experience depressive symptoms due to lower socioeconomic status, limited access to healthcare, and weaker social support networks. Self-assessed memory was also identified as an independent associated with factor, consistent with Xavier’s findings [37].
In addition, our study identified three cognitive function development trajectories among Chinese older adults. Tu et al. [38] analyzed the cognitive status of older adults in the Chinese Longitudinal Healthy Longevity Survey from 2002 to 2014, finding three trajectories: slow decline, rapid decline, and functional stabilization, which aligns with the present study’s findings. That study also examined factors associated with cognitive function trajectories. Age, education, gender, marital status, and ADL were found to be significantly associated with cognitive functioning, consistent with Yao’s research [39]. Furthermore, Zhang et al. [40] showed that older adults participating in social activities exhibited better cognitive function, supporting this study’s results. Social engagement may stimulate complex thinking and mental exercises, helping to delay cognitive decline [41]. Hearing status was also identified as an important factor associated with cognitive function. Brain changes, such as accelerated atrophy in areas essential for language and memory, are observed in individuals with hearing impairments [42], leading to cognitive decline. Additionally, health insurance can protect cognitive function by improving access to healthcare, supporting health management, and reducing financial stress.
Mild depressive symptoms are linked to better cognitive function, while high depressive symptoms are closely associated with cognitive decline. Conversely, cognitive function is also associated with depressive symptoms; higher cognitive functioning may be linked to lower occurrence or severity of depressive symptoms. Wang et al. [43] found a direct association between depressive symptoms and cognitive functioning through subgroup analyses and interactions using data from the 2011, 2013, and 2015 CHARLS. Yuan et al. [44] examined the interrelationships and heterogeneity in the progression of cognitive impairment and depressive symptoms in Chinese older adults by analyzing frailty, cognitive impairment, and depressive symptom trajectories. Zhang et al. [45] explored the link between depressive symptom status and cognitive decline, concluding that persistent depressive symptoms are associated with more rapid cognitive decline. Although these studies utilized long-term cohort data, they overlooked the variable heterogeneity, an important limitation that the GBDTM employed in the present study helps to address. The findings indicate that depressive symptoms and cognitive function are show reciprocal association among older adults. Therefore, in mental health assessments and interventions should take into account individuals’ cognitive functioning status and develop personalized treatment strategies tailored to different patterns of depressive symptoms to promote both mental and cognitive well-being.
Conclusion
This study analyzed 2,823 Chinese adults aged 60 years and older from CHARLS 2011 to 2018 and mapped heterogeneous longitudinal patterns of depressive symptoms and cognitive function. We identified three depressive symptom trajectories (low 31.0%, moderate 46.2%, high 22.8%) and three cognitive function trajectories (high 47.8%, declining 34.6%, low 17.6%). Depressive symptom groups mainly differed in overall symptom level with only small fluctuations across cycles, while cognitive function groups showed clearer separation in level and decline patterns over time. The dual trajectory results showed an inverse co development: the low depressive symptom group most often aligned with high cognitive function (0.61), and the high depressive symptom group most often aligned with cognitive decline (0.42). The low cognitive function group most often aligned with moderate depressive symptoms (0.49), and the joint distribution concentrated in combinations that paired higher cognitive function with milder depressive symptoms. We also found that higher education, better ADL function, and better self-rated health and memory related to more favorable trajectory membership. Together, these findings support psychological and aging research frameworks that treat affective and cognitive aging as connected processes and motivate integrated screening and targeted prevention that address both mood and cognitive vulnerability.
Limitations
Our study has several limitations in examining the relationship between depressive symptoms and cognitive function. Firstly, depressive symptoms were assessed using self-report scales, which may introduce information bias. Individuals’ subjective evaluations can be associated with their emotional state, comprehension, and differing interpretations of the questions, leading to measurement errors. The cognitive assessment tool used in this study was relatively limited. This constraint may limit the comprehensiveness of cognitive function evaluations. Future research should consider incorporating more diverse and multidimensional cognitive measures to improve the accuracy and generalizability of the findings. Secondly, although our findings summarize the temporal correspondence between depressive symptoms and cognitive function, the observational design does not allow for determination of the directionality or causal pathways of these associations, the complex interaction between the two requires further exploration. Furthermore, due to modeling convergence limitations, we were unable to identify demographic, health, and behavioral characteristics associated with the dual-trajectory patterns. Future studies should attempt to incorporate demographic, health, and behavioral factors into dual-trajectory models to gain deeper insight into how depressive symptoms and cognitive function in older adults are interrelated over time. Missing data may introduce bias. We deleted all observations with missing values in depressive symptoms, cognitive function, or any other model variables in the primary analysis, which may reduce power and may lead to selection bias if missingness relates to the outcomes. In sensitivity analyses, we still deleted participants with missing depressive symptoms or cognitive function because trajectory modeling depends on observed outcome information at each wave. When a wave has no observed items for an outcome scale, imputing the entire scale would rely heavily on information from other variables and other participants, which can add measurement error and affect trajectory identification. We therefore imputed missing values only for the other model variables using MICE and repeated the analyses. The overall patterns remained consistent, which supports robustness, although residual bias related to missing data may remain. Finally, this study adds evidence on heterogeneous longitudinal patterns of depressive symptoms and cognitive function in older adults. Because GBTM relies on repeated assessments, differential attrition and outcome-related nonresponse may affect the representativeness of the analytic sample and may limit generalizability to older adults who do not sustain follow-up participation. The identified trajectories reflect empirical summaries of longitudinal heterogeneity in depressive symptoms and cognitive function.
Supplementary Information
Acknowledgements
We are grateful for the data support provided by CHARLS, who provided valuable information for this study and facilitated the exploration of changes in cognitive function and depression symptoms and related risk factors.
Abbreviations
- WHO
World Health Organization
- MCI
Mild Cognitive Impairment
- GBTM
Group-Based Trajectory Modeling
- GBDTM
Group Based Dual-Trajectory Model
- BIC
Bayesian Information Criterion
- AIC
Akaike Information Criterion
- AvePP
average posterior probability
- CHARLS
China Health and Retirement Longitudinal Study
- CESD-10
10-item Center for Epidemiological Studies Depression scale
Authors’ contributions
ZXQ was responsible for drafting the initial manuscript, formulating the study concept, and collecting and analyzing the data. ZY and GRX carried out the implementation and conducted the feasibility assessment. ZHZ handled the statistical analyses, while DZW and JLH were in charge of chart creation and data visualization. BX and CSQ provided important revision suggestions, helping to improve the research content and clarity of expression, and LHY supervised the quality control process and provided critical revisions. JQ and ZWG offered methodological guidance and contributed to the overall design. All authors reviewed the manuscript.
Funding
The National Natural Science Foundation of China (72004165, 72374156) and the Team Research Project of the “Youth Innovation Team Program” at Shandong Provincial Higher Education Institution (2022RW075) provided funding for this work.
Data availability
The CHARLS was approved by the Ethics Committee of Peking University (IRB00001052-11015). The datasets analyzed during the current study are available in the CHARLS repository, https://opendata.pku.edu.cn.
Declarations
Ethics approval and consent to participate
This study was conducted in accordance with the ethical principles outlined in the Declaration of Helsinki. It used publicly available data from the China Health and Retirement Longitudinal Study (CHARLS), which obtained ethical approval from the Medical Ethics Review Committee of Peking University (IRB00001052-11015). All participants had provided written informed consent at the time of data collection.
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.
Contributor Information
Haiyan Li, Email: qdpdlhy@126.com.
Qi Jing, Email: jingq@sdsmu.edu.cn.
Wengui Zheng, Email: wgzheng@sdsmu.edu.cn.
References
- 1.Alexopoulos GS. Depression in the elderly. Lancet. 2005;365(9475):1961–70. [DOI] [PubMed] [Google Scholar]
- 2.Institute of Health Metrics and Evaluation. Global Health Data Exchange (GHDx). https://vizhub.healthdata.org/gbd-results/ Accessed 4 Mar. 2025.
- 3.China National Committee on Aging. China Aging Development Report 2024 – Mental Health of the Elderly in China. China Social Sciences Press. 2024. Accessed 4 Mar. 2025.
- 4.Kiely KM. Cognitive function. Encyclopedia of quality of life and well-being research. Dordrecht: Springer; 2014. pp. 974–8. [Google Scholar]
- 5.Jing, Qi. Lyu Jun. Community-based health service management strategies for elderly cognitive impairment [M]. Beijing: Science; 2024. [Google Scholar]
- 6.Alzheimer’s Disease International.World Alzheimer Report. 2023. Alzheimer’s Disease International, 2023. https://www.alz.co.uk/research/world-report-2023
- 7.Chen S, Chen X, Hou X, Fang H, Liu GG, Yan LL. Temporal trends and disparities of population attributable fractions of modifiable risk factors for dementia in China: a time-series study of the China health and retirement longitudinal study (2011–2018). Lancet Reg Health. 2024. 10.1016/j.lanwpc.2024.101106. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Ownby RL, Crocco E, Acevedo A, John V, Loewenstein D. Depression and risk for alzheimer disease: systematic review, meta-analysis, and metaregression analysis. Arch Gen Psychiatry. 2006;63(5):530–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Aajami Z, Toroski M, Bahrami M, Borhaninejad V. Relationship between depression and cognitive impairment among elderly: a cross-sectional study. J Caring Sci. 2020;9(3):148. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Zhu Y, Li C, Xie W, Zhong B, Wu Y, Blumenthal JA. Trajectories of depressive symptoms and subsequent cognitive decline in older adults: a pooled analysis of two longitudinal cohorts. Age Ageing. 2022;51(1):afab191. [DOI] [PubMed] [Google Scholar]
- 11.Han FF, Wang HX, Wu JJ, Yao W, Hao CF, Pei JJ. Depressive symptoms and cognitive impairment: a 10-year follow-up study from the Survey of Health, Ageing and Retirement in Europe. Eur Psychiatry. 2021;64(1):e55. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Vink D, Aartsen MJ, Schoevers RA. Risk factors for anxiety and depression in the elderly: a review. J Affect Disord. 2008;106(1–2):29–44. [DOI] [PubMed] [Google Scholar]
- 13.Cole MG, Dendukuri N. Risk factors for depression among elderly community subjects: a systematic review and meta-analysis. Am J Psychiatry. 2003;160(6):1147–56. [DOI] [PubMed] [Google Scholar]
- 14.Shin CH, Kim SY, Lee YS, Cho YC, Lee TY, Lee DB. A study on the factors related to the cognitive function and depression among the elderly. J Prev Med Public Health. 1996;29(2):199–214. [Google Scholar]
- 15.Clare, L., Wu, Y. T., Teale, J. C., MacLeod, C., Matthews, F., Brayne, C., … CFAS-Wales Study Team. (2017). Potentially modifiable lifestyle factors, cognitive reserve, and cognitive function in later life: A cross-sectional study. PLoS medicine, 14(3), e1002259. [DOI] [PMC free article] [PubMed]
- 16.Aichele S, Ghisletta P, Corley J, Pattie A, Taylor AM, Starr JM, Deary IJ. Fluid intelligence predicts change in depressive symptoms in later life: the Lothian birth cohort 1936. Psychol Sci. 2018;29(12):1984–95. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Graziane JA, Beer JC, Snitz BE, Chang CCH, Ganguli M. Dual trajectories of depression and cognition: a longitudinal population-based study. Am J Geriatr Psychiatry. 2016;24(5):364–73. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Zhang B, Lin Y, Hu M, Sun Y, Xu M, Hao J, et al. Associations between trajectories of depressive symptoms and rate of cognitive decline among Chinese middle-aged and older adults: an 8-year longitudinal study. J Psychosom Res. 2022;160:110986. [DOI] [PubMed] [Google Scholar]
- 19.Juan J, Yang H. Prevalence, prevention, and lifestyle intervention of gestational diabetes mellitus in China. Int J Environ Res Public Health. 2020;17(24):9517. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Gojnic M, Todorovic J, Stanisavljevic D, Jotic A, Lukic L, Milicic T, et al. Maternal and fetal outcomes among pregnant women with diabetes. Int J Environ Res Public Health. 2022;19(6):3684. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Nagin DS. (2009). Group-based trajectory modeling: an overview. Handbook of quantitative criminology, 53–67.
- 22.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]
- 23.Andresen EM, Malmgren JA, Carter WB, Patrick DL. Screening for depression in well older adults: evaluation of. Prev Med. 1994;10:77–84. [PubMed] [Google Scholar]
- 24.Cheng HG, Chen S, McBride O, Phillips MR. Prospective relationship of depressive symptoms, drinking, and tobacco smoking among middle-aged and elderly community-dwelling adults: results from the China health and retirement longitudinal study (CHARLS). J Affect Disord. 2016;195:136–43. [DOI] [PubMed] [Google Scholar]
- 25.Lei X, Hu Y, McArdle JJ, Smith JP, Zhao Y. Gender differences in cognition among older adults in China. J Hum Resour. 2012;47(4):951–71. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.McArdle JJ, Fisher GG, Kadlec KM. Latent variable analyses of age trends of cognition in the health and retirement Study, 1992–2004. Psychol Aging. 2007;22(3):525. [DOI] [PubMed] [Google Scholar]
- 27.Lin L, Cao B, Chen W, Li J, Zhang Y, Guo VY. Association of adverse childhood experiences and social isolation with later-life cognitive function among adults in China. JAMA Netw Open. 2022;5(11):e2241714–2241714. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Yu B, Steptoe A, Chen Y, Jia X. WITHDRAWN–Social isolation, rather than loneliness, is associated with cognitive decline in older adults: the China health and retirement longitudinal study. Psychol Med. 2024;54(15):4459–66. [DOI] [PubMed] [Google Scholar]
- 29.Li C, Jin S, Cao X, Han L, Sun N, Allore H, et al. Catastrophic health expenditure among Chinese adults living alone with cognitive impairment: findings from the CHARLS. BMC Geriatr. 2022;22(1):640. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Katz S, Ford AB, Moskowitz RW, Jackson BA, Jaffe MW. Studies of illness in the aged: the index of ADL: a standardized measure of biological and psychosocial function. JAMA. 1963;185(12):914–9. [DOI] [PubMed] [Google Scholar]
- 31.Wallace M. Katz index of independence in activities of daily living (ADL). Urol Nurs. 2007;27(1):93–4. [PubMed] [Google Scholar]
- 32.Nagin D. Group-based modeling of development Harvard university press. Cambridge: Mass; 2005. [Google Scholar]
- 33.Murphy RA, Hagaman AK, Reinders I, Steeves JA, Newman AB, Rubin SM, et al. Depressive trajectories and risk of disability and mortality in older adults: longitudinal findings from the health, aging, and body composition study. J Gerontol A Biol Sci Med Sci. 2016;71(2):228–35. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Akhtar-Danesh N, Landeen J. Relation between depression and sociodemographic factors. Int J Ment Health Syst. 2007;1:1–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Cui L, Ding D, Chen J, Wang M, He F, Yu S. Factors affecting the evolution of Chinese elderly depression: a cross-sectional study. BMC Geriatr. 2022;22(1):109. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Kao S, Wang YC, Tzeng YM, Liang CK, Lin FG. Interactive effect between depression and chronic medical conditions on fall risk in community-dwelling elders. Int Psychogeriatr. 2012;24(9):1409–18. [DOI] [PubMed] [Google Scholar]
- 37.Xavier FM, Ferraza MP, Argimon I, Trentini CM, Poyares D, Bertollucci PH, et al. The DSM-IV ‘minor depression’disorder in the oldest‐old: prevalence rate, sleep patterns, memory function and quality of life in elderly people of Italian descent in Southern Brazil. Int J Geriatr Psychiatry. 2002;17(2):107–16. [DOI] [PubMed] [Google Scholar]
- 38.Tu L, Lv X, Yuan C, Zhang M, Fan Z, Xu X, et al. Trajectories of cognitive function and their determinants in older people: 12 years of follow-up in the Chinese Longitudinal Healthy Longevity Survey. Int Psychogeriatr. 2020;32(6):765–75. [DOI] [PubMed] [Google Scholar]
- 39.Yao S, Zeng H, Sun S. Investigation on status and influential factors of cognitive function of the community-dwelling elderly in Changsha City. Arch Gerontol Geriatr. 2009;49(3):329–34. [DOI] [PubMed] [Google Scholar]
- 40.Fu C, Li Z, Mao Z. Association between social activities and cognitive function among the elderly in China: a cross-sectional study. Int J Environ Res Public Health. 2018;15(2):231. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Karp A, Paillard-Borg S, Wang HX, Silverstein M, Winblad B, Fratiglioni L. Mental, physical and social components in leisure activities equally contribute to decrease dementia risk. Dement Geriatr Cogn Disord. 2006;21(2):65–73. [DOI] [PubMed] [Google Scholar]
- 42.Lin FR, Albert M. Hearing loss and dementia–who is listening? Aging Ment Health. 2014;18(6):671–3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Wang J, Zhu A, Chen J, Zeng R, Wang S, Chen L, et al. Association between cognitive frailty and depression: a prospective cohort study of adults aged 45 years and older in China. Clin Interv Aging. 2024. 10.2147/cia.s484352. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Yuan Y, Peng C, Burr JA, Lapane KL. Frailty, cognitive impairment, and depressive symptoms in Chinese older adults: an eight-year multi-trajectory analysis. BMC Geriatr. 2023;23(1):843. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.Zhang N, Chao J, Cai R, Bao M, Chen H. The association between longitudinal changes in depressive symptoms and cognitive decline among middle-aged and older Chinese adults. Arch Gerontol Geriatr. 2023;109:104960. [DOI] [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 CHARLS was approved by the Ethics Committee of Peking University (IRB00001052-11015). The datasets analyzed during the current study are available in the CHARLS repository, https://opendata.pku.edu.cn.