Abstract
Objectives
As individuals age, cognitive function declines at varying degrees and rates. Cognitive reserve (CR) is thought to explain, to some extent, the individual differences observed in cognitive decline. However, evidence on the relationship between CR and cognitive trajectories remains limited. This study aimed to construct a proxy indicator of CR from a life course perspective and explore its relationship with cognitive trajectories.
Methods
Data were obtained from 3,460 older adults aged 60 and above, drawn from the 2011–2020 China Health and Aging Longitudinal Study (CHARLS). Proxy indicators for CR in early-life were sourced from the Life course panel, while middle and later-life indicators were derived from baseline data collected in 2011. Structural equation modeling (SEM) was employed to construct overall CR scores. Cognitive function was assessed through memory, orientation, and executive function, using group-based trajectory modeling (GBTM) to track cognitive aging trajectories over time. Logistic regression analysis was subsequently used to examine the association between CR and cognition over a 10-year follow-up period.
Results
Early CR’s most significant proxy indicators were education, type of health insurance in middle-life, and intellectual activity in later-life. Based on the final inclusion of 3,460 participants, three cognition trajectory groups were identified using GBTM: “Low-Rapid Decline” (23.0%), “Moderate-Gradual Decline” (40.3%), and “High-Stable” (36.7%). After adjusting for sociodemographic, lifestyle, and health variables, older adults with higher CR was associated with increased odds of being in the “High-Stable” cognition group (OR=1.9, 95% CI 1.74–2.07).
Conclusion
The study reveals that higher CR is more likely to be associated with more favorable cognitive trajectories, highlighting the cumulative life-course impact of CR on cognitive aging and robust longitudinal validity of its proxy indicators for promoting healthy aging worldwide.
Supplementary Information
The online version contains supplementary material available at 10.1186/s12889-025-24693-x.
Keywords: Cognitive reserve, Cognition, Life course perspective, Group-Based trajectory model, Older adults
Introduction
In the 21 st century, dementia has emerged as a significant challenge in public health and social care worldwide. By the end of 2022, the population of China aged 60 Years and older had exceeded 280 million, accounting for 19.8% of the total population [1]. As society ages rapidly, the incidence of cognitive disorders is increasing, imposing a growing burden on the economies and healthcare systems of various countries [2]. Given the lack of effective treatments for dementia, the development of prevention strategies has become particularly urgent.
Cognitive reserve (CR) refers to the brain’s adaptability in cognitive processes, such as efficiency, competence, and flexibility [3]. It helps explain why individuals vary in cognitive abilities and daily functioning despite brain aging, pathological changes, or disease [4]. CR is inherently a dynamic construct [5]. CR represents the brain’s functional capacity to maintain cognitive performance through compensatory neural networks and adaptive strategies [6]; CR should be differentiated from brain reserve, which denotes structural resilience at the neuroanatomical level, such as greater gray matter volume or synaptic density. Early hypotheses also propose that brain reserve enhances neuroplasticity by increasing the number of neurons or synapses and upregulating the expression of brain-derived neurotrophic factors, enabling the brain to withstand injury better [7]. It has been characterized as a “passive” form of reserve in that it does not actively combat the effects of pathology [8]. Brain maintenance refers to reduced development over time of age-related brain changes and pathology based on genetics or lifestyle [9–11].
Although CR cannot be measured directly, it is often inferred through multidimensional proxies such as education level, occupation, and social activities [12]. Some studies define CR as the discrepancy between actual cognitive function and the estimated function derived from overall AD neuropathology [13]. However, its core limitation lies in overly simplifying the complex neural compensation processes into a single pathology-cognition relationship.
Recent systematic reviews have expanded this study paradigm by defining CR evaluation methods and investigating seven modifiable lifestyle factors—diet, smoking, alcohol consumption, physical activity, cognitive leisure activity, sleep, and meditation, as potential modifiers of age/disease-related brain-cognition associations [14]. These studies provide evidence that lifestyle interventions may mitigate the impact of neurodegenerative pathology on cognitive decline, offering a dynamic perspective on CR beyond static pathological models. Studies further highlight CR’s role in delaying cognitive decline and dementia onset, positioning it as a promising target for preventive interventions [12, 15]. For example, higher educational attainment, occupational complexity, and intellectual leisure activities in late life consistently correlate with better cognitive function—factors widely used as CR proxies [16]. However, some studies have been limited in selecting of CR proxies, affecting their findings’ reliability.
This study innovatively constructs a proxy indicator system for CR using a life-course framework, Early-life stage reflects the foundation of neurodevelopmental resources [10]; Midlife stage represents the phase of proactive cognitive reserves [17]; Late-life stage embodies maintenance mechanisms [8].Compared to single time-point proxies, this multi-stage integration more closely aligns with the dynamic cumulative nature of CR. Moreover, the hierarchical construct validity of stage-specific indicators has been rigorously validated through structural equation modeling (SEM). Most previous longitudinal studies have employed linear mixed-effects models (LMMs) [16–18], which analyze individual differences in baseline cognitive scores and rates of cognitive decline by linking repeated measures to random effects. However, the model also has certain limitations. Specifically, LMMs presume a homogeneous population-wide cognitive trajectory, failing to capture potential heterogeneous subgroups with distinct decline patterns. This limitation necessitates alternative approaches like group-based trajectory modeling (GBTM) to identify discrete trajectory clusters. While cognitive trajectories can vary significantly across subjects, some individuals may share similar developmental trends, forming distinct cognitive trajectory groups. We hypothesized that higher cognitive reserve is associated with more favorable cognitive trajectories. Therefore, the present study will utilize a cluster trajectory model to assess the longitudinal associations between CR and cognitive trajectories over ten years.
Methods
Background of the study
The China Health and Retirement Longitudinal Study (CHARLS) is a large-scale, multidisciplinary social science research project initiated by the National Development Research Institute at Peking University, in collaboration with the China Social Science Survey Center [19]. The main objective of CHARLS is to collect comprehensive data on health, economic status, and retirement arrangements for individuals aged 45 and older in China, offering crucial insights into the aging population and providing data support for policy-making.
CHARLS uses a multi-stage stratified random sampling method and conducted its first survey in 2011–2012, covering 17,708 participants across 150 counties in 28 provinces. Follow-up surveys were conducted every two to three Years, with five rounds of data collected by 2020 (2011, 2013, 2015, 2018, and 2020). The project is ethically approved by the Biomedical Ethics Committee of Peking University, and informed consent was obtained from all participants [20]. For more detailed information on design sampling, and data collection, you can refer to prior publications or visit the official website: (http://www.icpsr.umich.edu/icpsrweb/NACDA/studies/36179).
Study population
In this study, we initially selected a sample of 4,213 participants aged 60 Years or older at baseline in 2011, all of whom had completed the Life-Course Survey in 2014, which provided data on early CR. Participants with complete baseline CR and cognitive data who had attended at least one follow-up visit were included. To avoid reverse causality, individuals with dementia, Parkinson’s disease, or other severe psychiatric disorders at baseline were excluded from the study. The final sample consisted of 3,460 participants. Figure S1 provides a detailed overview of the participant inclusion process.
CHARLS was approved by the Biomedical Ethics Review Board of Peking University (IRB00001052-11015). All participants signed an informed consent form before data collection.
Measurements
Assessment of cognition
Cognitive functioning was assessed every two Years using a standardized questionnaire, including memory, orientation, and executive function measures. Memory function was evaluated through immediate and delayed word recall tests. The orientation function was assessed by asking participants to identify the Year, month, day, day of the week, and season at the testing time. Executive function was measured using the serial subtraction of 7s from 100 (100-7). For all tests, higher scores indicated better cognitive performance. Overall cognitive functioning in this study was defined as the sum of scores across the three dimensions. Previous studies have confirmed the validity and reliability of these tests [21, 22]. Detailed information on the cognitive function tests can be found in Appendix Table S1.
Assessment of cognitive reserve
This study explored the contributions of life course stages to CR, with a specific focus on early, middle, and later- life. CR was evaluated based on participants’ living conditions during childhood, socioeconomic status in middle-life, and social networks and leisure activities in later-life. Early-life factors were defined by family economic status, parents’ education level and occupation, and nutritional and dietary conditions, with data primarily sourced from the 2014 Life Course Survey [23]. CR in middle-life was measured through indicators such as household income, education level, occupation type, and health insurance coverage [24, 25]. Later-life CR was assessed through marital status, place of residence, number of children, frequency of contact with family members, and participation in social and leisure activities. Detailed CR indicator measures are summarized in Appendix Table S1.
The raw scores for the three life stages were converted into Z-scores using the baseline mean and standard deviation. These Z-scores were then summed to obtain a composite CR score. The final composite CR score was calculated by summing the three standardized metrics, each weighted according to its contribution.
Although principal component factor analysis is commonly used to explain the full variance or covariance of a set of indicator variables, structural equation modeling (SEM) allows for retaining of each indicator’s unique variance [17]. Therefore, this study employed second-order confirmatory factor analysis (CFA) within SEM, using the maximum likelihood estimation method to evaluate CR more comprehensively. Model fit was evaluated using goodness-of-fit metrics, including the chi-square to degrees of freedom ratio (χ²/df), root-mean-square error of approximation (RMSEA), and Comparative Fit Index (CFI) [26], as detailed in Appendix Table S2. The final modeling process is presented in Fig. 1. Additionally, we have redrawn a Table S3 in Appendix to provide a detailed explanation for the inclusion of the proxy indicators of CR in the structural equation model. We also explain why some low loading factors are still retained in the structural equation model.
Fig. 1.
Standardized estimates for the composite cognitive reserve score derived from the structural equation models (n = 3460). The values indicate the β-coefficients of the four observable factors over the life course used to generate the composite score of cognitive reserve from the structural equation models. . e1 ~ e19 represent the measurement error for each of the four observable factors in estimating the composite cognitive reserve score. Estimator: Maximum Likelihood Robust. The values indicate the loadings of the markers to first-order factors and to cognitive reserve
Covariates
This study included appropriate covariates based on previous research and univariate regression analyses [27–29]. The multivariate adjusted model accounted for the following covariates: age (continuous), sex (male, female), residence (urban, rural), smoking (never, former, current), drinking (never, former, current), body mass index (BMI: <18.5, 18.5–23.9, 24.0-27.9, ≥ 28.0 kg/m²), self-rated health (excellent, good, fair, poor, very poor), comorbidities (0, 1, ≥ 2), depressive symptoms (yes, no), visual impairment (yes, no), hearing impairment (yes, no), hospitalization (yes, no), sleep disorders (yes, no), and walking time (continuous).
Walking time was defined as the average time taken for participants to walk twice along a 2.5-meter Linear path. Depressive symptoms were assessed using the 10-item version of the Center for Epidemiological Studies Depression Scale (CES-D), with questions 5 and 8 reverse-coded. Each item was scored on a 4-point scale (0 to 3), with total scores ranging from 0 to 30, reflecting increasing severity of depression. A score of ≥ 12 indicated the presence of depressive symptoms [18].
Functional status in basic and instrumental activities of daily living (ADL) was assessed using an established scale. Responses indicating normal functioning were coded as 0, while those indicating functional impairment were coded as 1. Participants were classified as having 0, 1, or 2 + functional impairments [30].
Body mass index (BMI) was calculated as weight (in kilograms) divided by height (in meters squared) and classified into four categories: underweight (< 18.5 kg/m²), normal (18.5–23.9 kg/m²), overweight (24.0–27.9 kg/m²), and obese (≥ 28.0 kg/m²). Missing data were addressed using multiple imputation techniques to ensure a complete dataset for analysis.
Statistical analyses
We used group-based trajectory modeling (GBTM) to identify distinct trajectories of cognition and activity scores across each wave. GBTM allows for the inclusion of all available scores in the model estimates, assuming that any missing data occur at random. The successive Z-scores were modeled as censored normal averages. A maximum of six trajectory groups was set a priori based on initial analyses. Models ranging from one to six trajectory groups were fitted, and the optimal number of trajectories was determined by first applying the cubic model and then adjusting according to the p-value. The Bayesian Information Criterion (BIC) was used to identify the best-fitting model. Additionally, the average posterior probability (AvePP) of assigning each participant to a trajectory group was used as a criterion for model fit, with an AvePP of approximately 70% or higher indicating a good fit. The model’s accuracy was further evaluated using the Odds of Correct Classification (OCC), with an OCC above five considered highly accurate for all trajectory groups. Models with more than 10% of participants assigned to each trajectory group were selected [31–33].
Descriptive statistics are reported as n (%) for categorical variables and mean ± standard deviation (SD) for continuous variables. Baseline differences across tertiles were compared using analysis of variance (ANOVA) for continuous variables and chi-square tests for categorical variables. Both univariate and multivariate logistic regression analyses were performed. Multivariate models were employed to estimate associations between cognitive function trajectories and CR, reporting odds ratios (ORs) and their corresponding 95% confidence intervals (CIs).
The relationship between early, middle, late, and overall CR and cognitive function was initially analyzed using unadjusted models. Covariates were then adjusted in two additional models. Model 1 controlled for age, sex, and place of residence, and Model 2 further adjusted for lifestyle factors (smoking, alcohol consumption, sleep quality, and BMI) and clinical factors (self-rated health, comorbidities, hearing and visual impairments, depressive symptoms, hospitalization, and walking time).
In the sensitivity analysis, the study first performed stratified analyses by age group (< 65 and ≥ 65 years), sex (male and female), depression status (yes or no), and ADL functional status (0, 1, ≥ 2), and examined the interaction effect between total CR and cognitive function trajectory groups. Second, baseline characteristics of individuals included in the study were compared with those of individuals lost to follow-up. Third, as trajectory analyses are considered more stable for participants with three or more observations, regression analyses were rerun for participants with cognitive function measures from all three waves. Finally, to minimize confounding effects from CR proxy choices, cognitive reserve dimensions were regenerated as scores based solely on observed variables, without multiplying by loading coefficients.
All statistical analyses were conducted using R packages (http://www.R-project.org, The R Foundation) with PROC TRAJ and Free Statistics software version 1.9.2. All statistical tests were two-sided, and a p-value of < 0.05 was considered statistically significant.
Results
Characteristics of the participants
Table 1 outlines the characteristics of the study population (n = 3,460). The mean age of participants was 66.6 ± 5.5 Years, with 57.3% male and 72.9% residing in rural areas. A majority of participants (84.3%) were married, and 46.0% had an educational attainment of elementary school or below. The baseline cognitive score for the entire population was 11.9 ± 3.1. Specifically, participants in the low CR group had a mean score of 10.3 ± 2.9, those in the middle CR group scored 11.3 ± 3.0, and individuals in the high CR group scored 13.0 ± 2.8.
Table 1.
Baseline characteristics of the participants according to per tertile of the cognitive reserve. (n = 3,460)
| Variables | Total (n = 3460) |
Low (n = 714) |
Medium (n = 1176) |
High (n = 1570) |
P |
|---|---|---|---|---|---|
| Age, Mean ± SD | 66.6 ± 5.5 | 66.5 ± 5.7 | 66.3 ± 5.3 | 66.8 ± 5.5 | 0.043 |
| Male, n (%) | 1984 (57.3) | 366 (51.3) | 639 (54.3) | 979 (62.4) | < 0.001 |
| Rural residence, n (%) | 2524 (72.9) | 671 (94) | 1054 (89.6) | 799 (50.9) | < 0.001 |
| Marital status, n (%) | < 0.001 | ||||
| Never Married | 24 (0.7) | 24 (3.4) | 0 (0) | 0 (0) | |
| Widowed | 477 (13.8) | 220 (30.8) | 149 (12.7) | 108 (6.9) | |
| Divorced/Separated | 44 (1.3) | 21 (2.9) | 10 (0.9) | 13 (0.8) | |
| Married | 2915 (84.3) | 449 (62.3) | 37 (86.5) | 1449 (92.3) | |
| Educational level, n (%) | < 0.001 | ||||
| No formal education | 786 (22.7) | 333 (46.6) | 306 (26) | 147 (9.4) | |
| Primary school | 805 (23.3) | 243 (34) | 326 (27.7) | 236 (15) | |
| Middle school | 1043 (30.1) | 124 (17.4) | 423 (36) | 496 (31.6) | |
| high school | 552 (16.0) | 14 (2) | 109 (9.3) | 429 (27.3) | |
| College or above | 274 (7.9) | 0 (0) | 12 (1) | 262 (16.7) | |
| Self-report health, n (%) | < 0.001 | ||||
| Very good | 143 (4.1) | 44 (6.2) | 58 (4.9) | 41 (2.6) | |
| Good | 813 (23.5) | 199 (27.9) | 311 (26.4) | 303 (19.3) | |
| Fair | 1774 (51.3) | 332 (46.5) | 584 (49.7) | 858 (54.6) | |
| Poor | 539 (15.6) | 106 (14.8) | 164 (13.9) | 269 (17.1) | |
| Very poor | 191 (5.5) | 33 (4.6) | 59 (5) | 99 (6.3) | |
| BMI (kg/m2), n(%)a | < 0.001 | ||||
| Underweight | 271 (7.8) | 68 (9.5) | 113 (9.6) | 90 (5.7) | |
| Normal | 1891 (54.7) | 440 (61.6) | 655 (55.7) | 796 (50.7) | |
| Overweight | 959 (27.7) | 152 (21.3) | 313 (26.6) | 494 (31.5) | |
| Obese | 339 (9.8) | 54 (7.6) | 95 (8.1) | 190 (12.1) | |
| Household income, n (%) | < 0.001 | ||||
| Very low | 575 (16.6) | 285 (39.9) | 220 (18.7) | 70 (4.5) | |
| Low | 669 (19.3) | 200 (28) | 340 (28.9) | 129 (8.2) | |
| Middle | 720 (20.8) | 151 (21.1) | 308 (26.2) | 261 (16.6) | |
| High | 748 (21.6) | 61 (8.5) | 191 (16.2) | 496 (31.6) | |
| Very high | 748 (21.6) | 17 (2.4) | 117 (9.9) | 614 (39.1) | |
| Depressive symptoms, n (%) | 813 (23.5) | 240 (33.6) | 322 (27.4) | 251 (16) | < 0.001 |
| Comorbidity, n(%)b | < 0.001 | ||||
| 0 | 857 (24.8) | 208 (29.1) | 304 (25.9) | 345 (22) | |
| 1 | 972 (28.1) | 192 (26.9) | 335 (28.5) | 445 (28.3) | |
| ≥ 2 | 1631 (47.1) | 314 (44) | 537 (45.7) | 780 (49.7) | |
| ADL, n (%) | < 0.001 | ||||
| 0 | 2812 (81.3) | 535 (74.9) | 925 (78.7) | 1352 (86.1) | |
| 1 | 356 (10.3) | 92 (12.9) | 130 (11.1) | 134 (8.5) | |
| ≥ 2 | 292 (8.4) | 87 (12.2) | 121 (10.3) | 84 (5.4) | |
| Drinking, n (%) | 0.563 | ||||
| Never | 1910 (55.2) | 408 (57.1) | 651 (55.4) | 851 (54.2) | |
| Former | 404 (11.7) | 74 (10.4) | 145 (12.3) | 185 (11.8) | |
| Current | 1146 (33.1) | 232 (32.5) | 380 (32.3) | 534 (34) | |
| Smoking, n (%) | < 0.010 | ||||
| Never | 1816 (52.5) | 376 (52.7) | 638 (54.3) | 802 (51.1) | |
| Former | 440 (12.7) | 72 (10.1) | 136 (11.6) | 232 (14.8) | |
| Current | 1204 (34.8) | 266 (37.3) | 402 (34.2) | 536 (34.1) | |
| Visual impairment, n (%) | 562 (16.2) | 78 (10.9) | 160 (13.6) | 324 (20.6) | < 0.001 |
| Hearing impairment, n (%) | 527 (15.2) | 130 (18.2) | 191 (16.2) | 206 (13.1) | < 0.001 |
| Sleep disorder, n (%) | 1731 (50.0) | 393 (55) | 615 (52.3) | 723 (46.1) | < 0.001 |
| Hospitalization, n (%) c | 359 (10.4) | 66 (9.2) | 107 (9.1) | 186 (11.8) | 0.035 |
| Total Cognition, Mean ± SD | 11.9 ± 3.1 | 10.3 ± 2.9 | 11.3 ± 3.0 | 13.0 ± 2.8 | < 0.001 |
| Gait time (S/2.5 m), Mean ± SD d | 1.8 ± 1.1 | 1.7 ± 0.8 | 1.8 ± 1.0 | 1.9 ± 1.3 | < 0.001 |
SD, Standard Deviation
aBMI, body mass index
bComorbidity,Comorbidity refers to the coexistence of multiple chronic conditions, such as hypertension, diabetes, dyslipidemia, and chronic kidney disease, along with other chronic illnesses diagnosed by physicians, including heart disease, stroke, lung disease, arthritis, and cancer. It is quantified It is quantified based on the number of these nine chronic diseases present in an individual, categorized as none (0), one (1), or multiple (≥ 2)
cHospitalization, Hospitalization refers to whether an individual was hospitalized during the past year
dGait time, Time taken by participants to walk 2.5 meters
In addition to cognitive scores, participants in the lowest CR tertile exhibited lower educational attainment, poorer self-reported health, and lower income compared to those in the highest CR tertile. They also demonstrated a higher prevalence of sleep disorders and slower gait speeds. Further details are provided in Table 1.
The cognitive reserve model, constructed using structural equation modeling (SEM), is depicted in Fig. 1. The highest loading factor for early-life CR was education level (0.66), for middle-life CR it was type of health insurance (0.76), and for later-life CR it was participation in intellectual activities (0.28).
Cognitive trajectory group characteristics
Cognitive trajectories for groups 2–6 were fitted to identify the optimal model (Appendix Tables S4). Although the four-group model had the lowest BIC in the cognitive trajectory group, the mean posterior probability was less than 0.7, indicating a poor fit. Consequently, a three-trajectory model was selected as the optimal solution. Figure 2 illustrates the three cognitive trajectory patterns identified in the study: Class 1, “Low-Rapid Decline Cognition” (N = 796, 23.0%), Class 2, “Moderate-Gradual Decline Cognition” (N = 1,394, 40.3%), and Class 3, “High-Stable Cognition” (N = 1,270, 36.7%).
Fig. 2.
Cognitive trajectory group. The horizontal axis represents time, and the vertical axis represents the average cognitive scores of the trajectory groups
Table 2 presents the baseline characteristics of participants in each cognitive trajectory group. Compared to those in the Low-Rapid Decline group, the High-Stable Cognition group participants were more likely to be younger, male, have higher educational attainment, reside in urban areas, and report higher incomes. Additionally, they exhibited better overall health, a more remarkable ability to perform daily activities, and a lower prevalence of depressive symptoms, as well as visual and auditory impairments.
Table 2.
Association between demographic and clinical characteristics at admission and cognitive impairment trajectory groupsa
| Cognitive function trajectories | ||||
|---|---|---|---|---|
| Variables | (ref: Low-Rapid Decline cognition) | |||
| Moderate-Gradual Decline | High-Stable | |||
| OR (95% CI) | P-value | OR (95% CI) | P-value | |
| Age | 0.95 (0.94 ~ 0.97) | < 0.001 | 0.91 (0.89 ~ 0.92) | < 0.001 |
| Sex | ||||
| Female | 1 (Ref) | 1 (Ref) | ||
| Male | 2.1 (1.76 ~ 2.5) | < 0.001 | 2.7 (2.25 ~ 3.24) | < 0.001 |
| Residence | ||||
| Urban | 1 (Ref) | 1 (Ref) | ||
| Rural | 0.45 (0.35 ~ 0.59) | < 0.001 | 0.14 (0.11 ~ 0.18) | < 0.001 |
| Marital status | ||||
| Never Married | 1 (Ref) | 1 (Ref) | ||
| Widowed | 0.5 (0.19 ~ 1.31) | 0.157 | 1.56 (0.38 ~ 6.35) | 0.537 |
| Divorced/Separated | 0.68 (0.2 ~ 2.32) | 0.538 | 3.4 (0.69 ~ 16.69) | 0.132 |
| Married | 0.78 (0.3 ~ 2.02) | 0.607 | 3.71 (0.93 ~ 14.9) | 0.064 |
| Educational level | ||||
| No formal education | 1 (Ref) | 1 (Ref) | ||
| Primary school | 3.88 (3.1–4.87) | < 0.001 | 9.46 (6.57 ~ 13.62) | < 0.001 |
| Middle school | 8.96 (6.87 ~ 11.67) | < 0.001 | 47.78 (32.86–69.49) | < 0.001 |
| high school | 11.22 (7.17–17.54) | < 0.001 | 148.87 (89.76 ~ 246.88) | < 0.001 |
| College or above | 21.2 (7.57–59.4) | < 0.001 | 563.3 (200.26 ~ 1584.47) | < 0.001 |
| Self-report health | ||||
| Very good | 1 (Ref) | 1 (Ref) | ||
| Good | 1.32 (0.88 ~ 1.98) | 0.186 | 1.55 (0.96 ~ 2.49) | 0.075 |
| Fair | 1.59 (1.07 ~ 2.36) | 0.021 | 2.86 (1.81 ~ 4.54) | < 0.001 |
| Poor | 2 (1.28 ~ 3.13) | 0.002 | 4.42 (2.67 ~ 7.32) | < 0.001 |
| Very poor | 1.75 (1.02 ~ 3) | 0.044 | 3.36 (1.86 ~ 6.06) | < 0.001 |
| Living alone | ||||
| No | 1 (Ref) | 1 (Ref) | ||
| Yes | 1.58 (1.25 ~ 1.99) | < 0.001 | 1.74 (1.36 ~ 2.21) | < 0.001 |
| BMI | ||||
| Underweight | 1 (Ref) | 1 (Ref) | ||
| Normal | 1.13 (0.84 ~ 1.53) | 0.417 | 1.92 (1.35 ~ 2.74) | < 0.001 |
| Overweight | 1.37 (0.99 ~ 1.91) | 0.061 | 3.24 (2.23 ~ 4.72) | < 0.001 |
| Obese | 1.15 (0.77 ~ 1.71) | 0.503 | 2.57 (1.66 ~ 3.98) | < 0.001 |
| Annual household income | ||||
| Very low | 1 (Ref) | 1 (Ref) | ||
| Low | 1.14 (0.88 ~ 1.48) | 0.316 | 1.2 (0.88 ~ 1.62) | 0.246 |
| Middle | 1.31 (1.01 ~ 1.71) | 0.043 | 1.89 (1.41 ~ 2.55) | < 0.001 |
| High | 1.62 (1.23 ~ 2.14) | 0.001 | 3.3 (2.45 ~ 4.47) | < 0.001 |
| Very high | 1.64 (1.21 ~ 2.21) | 0.001 | 5.55 (4.07 ~ 7.57) | < 0.001 |
| Depressive Symptoms | ||||
| No | 1 (Ref) | 1 (Ref) | ||
| Yes | 0.61 (0.51 ~ 0.74) | < 0.001 | 0.32 (0.26 ~ 0.4) | < 0.001 |
| Comorbidity | ||||
| 0 | 1 (Ref) | 1 (Ref) | ||
| 1 | 1.13 (0.89 ~ 1.44) | 0.298 | 1.14 (0.89 ~ 1.45) | 0.302 |
| ≥ 2 | 1.11 (0.89 ~ 1.37) | 0.354 | 1.18 (0.95 ~ 1.47) | 0.133 |
| ADL | ||||
| 0 | 1 (Ref) | 1 (Ref) | ||
| 1 | 0.71 (0.55 ~ 0.93) | 0.013 | 0.44 (0.33 ~ 0.6) | < 0.001 |
| ≥ 2 | 0.49 (0.37 ~ 0.64) | < 0.001 | 0.24 (0.18 ~ 0.34) | < 0.001 |
| Drinking | ||||
| Never | 1 (Ref) | 1 (Ref) | ||
| Former | 1.3 (0.98 ~ 1.72) | 0.066 | 1.12 (0.84 ~ 1.5) | 0.441 |
| Current | 1.52 (1.25 ~ 1.85) | < 0.001 | 1.54 (1.26 ~ 1.89) | < 0.001 |
| Smoking | ||||
| Never | 1 (Ref) | 1 (Ref) | ||
| Former | 1.69 (1.26 ~ 2.27) | < 0.001 | 1.81 (1.34 ~ 2.43) | < 0.001 |
| Current | 1.47 (1.21 ~ 1.78) | < 0.001 | 1.38 (1.13 ~ 1.68) | 0.001 |
| Visual impairment | ||||
| No | 1 (Ref) | 1 (Ref) | ||
| Yes | 1.42 (1.09 ~ 1.86) | 0.01 | 2.18 (1.68 ~ 2.83) | < 0.001 |
| Hearing impairment | ||||
| No | 1 (Ref) | 1 (Ref) | ||
| Yes | 0.78 (0.62 ~ 0.98) | 0.03 | 0.52 (0.4 ~ 0.66) | < 0.001 |
| Sleep disorder | ||||
| No | 1 (Ref) | 1 (Ref) | ||
| Yes | 0.85 (0.72 ~ 1.02) | 0.074 | 0.65 (0.55 ~ 0.78) | < 0.001 |
| Hospitalization | ||||
| No | 1 (Ref) | 1 (Ref) | ||
| Yes | 0.74 (0.56 ~ 0.98) | 0.036 | 0.83 (0.63 ~ 1.1) | 0.205 |
| Gait time (S/2.5 m) | 0.92 (0.86 ~ 0.99) | 0.021 | 0.87 (0.81 ~ 0.94) | < 0.001 |
Notes: CI: confidence interval
aMultiple imputation results are summarized from 5 imputed data sets
bBMI, body mass index
Associations of CR with cognitive function
Table 3 summarizes the results of the multinomial regression analyses examining CR across early, middle, and later-life and over the Life course. CR was included in the models as both tertile and continuous variables. After adjustment, individuals with higher cognitive reserve across all Life stages demonstrated better cognitive trajectories than those with lower cognitive reserve. The most substantial effect sizes were observed in early-life, where individuals in Tertile 3 had a significantly higher likelihood of being in the High Cognitive Functioning Trajectory group, with odds ratios (ORs) and 95% confidence intervals (CIs) ranging from 15.89 (11.93 to 21.18). Similar results were found for CR as a continuous variable.
Table 3.
Multivariate logistic regression analyses with cognitive function trajectory groups
| Variables | Crude model | Model 1 | Model 2 | |||
|---|---|---|---|---|---|---|
| Moderate-Gradual Decline | High-Stable | Moderate-Gradual Decline | High-Stable | Moderate-Gradual Decline | High-Stable | |
| Early-life | ||||||
| Tertile 1 | 1 (Ref) | 1 (Ref) | 1 (Ref) | 1 (Ref) | 1 (Ref) | 1 (Ref) |
| Tertile 2 |
2.34 (1.86–2.95)*** |
5.85 (4.44–7.7)*** |
1.11 (0.89 ~ 1.4) |
2.47 (1.86–3.27)*** |
2.08 (1.64–2.64)*** |
4.79 (3.56–6.46)*** |
| Tertile 3 |
4.79 (3.81–6.02)*** |
24.1 (18.55–31.3)*** |
1.47 (1.02–2.12)* |
3.39 (2.23–5.16)*** |
4.1 (3.22 ~ 5.22)*** |
15.89 (11.93–21.18)*** |
| Continuous |
2.95 (2.56–3.4)*** |
6.72 (5.76–7.84)*** |
1.11 (0.93 ~ 1.33) |
1.27 (1.04–1.56)* |
2.67 (2.3–3.1)*** |
5.37 (4.55–6.35)*** |
| Middle-life | ||||||
| Tertile 1 | 1 (Ref) | 1 (Ref) | 1 (Ref) | 1 (Ref) | 1 (Ref) | 1 (Ref) |
| Tertile 2 |
1.34 (1.1–1.64)** |
2.27 (1.78–2.9)*** |
0.92 (0.71 ~ 1.18) |
1.09 (0.78 ~ 1.52) |
1.09 (0.88 ~ 1.35) |
1.57 (1.2–2.04)*** |
| Tertile 3 |
2.18 (1.68–2.84)*** |
9.59 (7.23–12.73)*** |
0.88 (0.59 ~ 1.32) |
1.09 (0.66 ~ 1.82) |
1.22 (0.9 ~ 1.66) |
2.67 (1.89–3.75)*** |
| Continuous |
1.46 (1.31–1.62)*** |
2.9 (2.59–3.25)*** |
1.29 (1.04–1.61)* |
2.32 (1.8–2.97)*** |
1.16 (1.02–1.32)* |
1.68 (1.45–1.94)*** |
| Later-life | ||||||
| Tertile 1 | 1 (Ref) | 1 (Ref) | 1 (Ref) | 1 (Ref) | 1 (Ref) | 1 (Ref) |
| Tertile 2 |
0.78 (0.64–0.97)* |
0.73 (0.59–0.9)** |
0.87 (0.75 ~ 1.02) |
1.06 (0.89 ~ 1.28) |
0.94 (0.76 ~ 1.17) |
0.99 (0.78 ~ 1.27) |
| Tertile 3 |
0.93 (0.74 ~ 1.16) |
1.14 (0.91 ~ 1.42) |
0.98 (0.83 ~ 1.16) |
1.55 (1.28 ~ 1.89) |
1.1 (0.87 ~ 1.4) |
1.48 (1.14–1.92)** |
| Continuous |
0.93 (0.85 ~ 1.02) |
1.02 (0.93 ~ 1.12) |
0.99 (0.92 ~ 1.06) |
1.21 (1.11 ~ 1.31)*** |
1 (0.9 ~ 1.11) |
1.14 (1.02–1.28)* |
| Total CR | ||||||
| Tertile 1 | 1 (Ref) | 1 (Ref) | 1 (Ref) | 1 (Ref) | 1 (Ref) | 1 (Ref) |
| Tertile 2 |
1.63 (1.32 ~ 2.01) *** |
2.85 (2.18 ~ 3.74) *** |
1.24 (1 ~ 1.55)* |
1.18 (0.88 ~ 1.58) |
1.61 (1.29–2.01)*** |
2.76 (2.06–3.7)*** |
| Tertile 3 |
3.03 (2.4–3.82)*** |
12.97 (9.86–17.06)*** |
1.73 (1.18–2.53)** |
1.14 (0.72 ~ 1.81) |
2.55 (1.97–3.3)*** |
7.58 (5.55–10.35)*** |
| Continuous |
1.39 (1.3–1.49)*** |
2.13 (1.98–2.28)*** |
1.02 (0.92 ~ 1.13) |
1.35 (1.21–1.52)*** |
1.33 (1.24–1.44)*** |
1.9 (1.74–2.07)*** |
Model 1: adjusted for age, gender, residence
Model 2: additionally adjusted for smoking and drinking, sleep, BMI, Self-Rated Health Status, Comorbidities, Hearing and Vision, Depression Symptoms, Hospitalization, and Gait time. Data are OR [95% CI]; ref: Low-Rapid Decline cognition
*P < 0.05
**P < 0.01
***P < 0.001
However, in later-life, the association between higher CR and the maintenance of high cognitive functioning was less pronounced than in the early-life, with many ORs approaching 1. This suggests that CR in later-life contributes less to maintaining cognitive function compared to early and middle-life. These findings are visually depicted in Table 3.
A significant interaction was found between the composite CR score and depression when comparing the " Low-Rapid Decline " and " High-stable” cognitive trajectory groups (interaction effect P = 0.03). However, the association between cognitive reserve score and cognitive trajectory group was consistent across different subgroups, including age, gender, ADL status, and rural versus urban residency (all interaction effect P values > 0.1). Further details are provided in Appendix Figure S3.
First, the present study compared baseline characteristics between individuals who completed the baseline survey (n = 3,674) and those lost to follow-up (n = 214). The results indicated that the 214 individuals lost to follow-up (6.2%) had higher levels of significant risk factors but exhibited good cognitive function at baseline (Additional Table S5).
Given that trajectory analyses are more stable for participants with three or more observations [34], additional regression analyses were conducted for participants with cognitive function measurements across all three waves (n = 2,941). The cognitive trajectory patterns using this complete data were consistent with the primary analysis: Class 1, “Low-Rapid Decline in Cognition” (n = 591, 20.1%), Class 2, “Moderate-Gradual Decline in Cognition” (n = 1,114, 37.9%), and Class 3, “High-Stable Cognition” (n = 1,138, 38.7%) (Additional File 1: Figure S2). The associations between CR, total CR, and cognitive trajectory scores at each life stage were consistent with the primary analysis (Additional Table S6-S7).
Additionally, regression analyses were re-run using only the raw scores of the observed CR. The associations between CR at each life stage and overall cognitive trajectories remained similar to the primary analyses. The robustness of the study’s findings was confirmed through multiple sensitivity tests.
Discussion
This study identified three distinct cognitive trajectory groups over a 10-year follow-up period. The three cognitive trajectory patterns identified were “Low-Rapid Decline Cognition,” “Moderate-Gradual Decline Cognition,” and “High-Stable Cognition.” CR was found to be associated with cognitive functioning in later-life, and distinct cognitive trajectories observed among different CR populations. The contribution of CR proxies varied across three time periods, with education, health insurance, and intellectual activity being the most significant proxies.
We found that most older adults exhibited a relatively stable cognitive trend over 10 Years, which is consistent with the conclusions of a 6-month cognitive trajectory study in older adults [35]. However, other studies have classified cognitive trajectories into “average-stable,” “high-stable,” and “declining” patterns, with the declining proportion accounting for less than 10% [36]. However, in our study, with the passage of time, the 2018 survey showed that the “Low-Rapid Decline Cognition” and “Moderate-Gradual Decline Cognition” groups experienced some cognitive improvement. This may be related to the promotion period of China’s “Healthy Aging” policy during the study period [37], when cognitive health promotion activities for older adults were likely popularized in the sampled communities. Additionally, social environmental changes, such as home quarantine during the pandemic, may have promoted intergenerational family interactions, indirectly enhancing older adults’ cognitive [38]. The high-cognition group remained consistently stable, possibly because individuals with high cognitive function possess stronger CR. Even in the presence of age-related neurodegenerative changes or life event shocks, their brains can maintain functional stability, leading to no significant fluctuations in explicit cognitive performance. These results align with previous studies [12]. While earlier studies have also shown that higher CR is associated with faster cognitive decline after a certain age, particularly in extremely aged populations [17]. However, this was not observed in our study. One possible explanation may be the relatively young average age of the participants. Additionally, most previous studies employed different tools to assess CR and cognitive functioning, particularly regarding CR proxies.
This study demonstrates that CR-related proxies positively correlate with cognitive function. Given that CR develops throughout life, early investigation of life experiences, midlife socioeconomic status, and later-life social activities can aid in identifying individuals at risk for cognitive decline. CFA identified early education as a critical variable influencing factor loadings. Education level has also been a widely used proxy in past research [15]. In China, low education is one of the leading modifiable risk factors for dementia [39], with approximately 10.8% of cases attributed to insufficient education [40]. Individuals with lower education levels or poor early academic performance may develop less efficient or flexible neural networks, making them more vulnerable to dementia [41]. Education also mitigates cognitive decline through various pathways. For instance, higher education often correlates with better income, which allows access to quality healthcare. Individuals with higher education levels are more likely to engage in cognitively demanding jobs, and prioritize physical activity and healthcare, all contributing to slowing cognitive decline. Research highlights the importance of addressing low education levels as a critical dementia risk factor in China [42]. Those with better health insurance benefit from regular check-ups, early disease screening, and timely treatments, which enhance access to high-quality healthcare [25]. These factors enable the early detection and management of health issues that may impair cognitive function. Additionally, individuals in cognitively demanding jobs regularly engage in information processing, problem-solving, and decision-making. Long-term cognitive stimulation from such activities enhances neuroplasticity, the brain’s capacity to adapt and repair [29].
Factor loadings for intellectual activities in later-life outperform those for other social activities. This is consistent with previous studies showing that older adults engaging in intellectual activities have better cognitive performance. However, caution is needed when interpreting these findings. Furthermore, given the potential for reverse causality, where cognitive decline might reduce physical and social activities [43], further research is needed to explore these complex relationships.
This study found that early-life CR had a much more substantial impact on cognitive outcomes in the better trajectory group than the composite CR indicator in later-life. Specifically, early-life CR provides a significant and lasting protective effect among older Chinese older adults, mainly due to the foundational role of early education and cognitive activities. In contrast, later-life CR is primarily influenced by socialization opportunities, often limited in rural and economically underdeveloped areas of China [44]. The widespread phenomenon of “empty-nest” families, changes in family structures during urbanization, and the resulting social isolation have deprived many older adults of maintaining social networks. This lack of social interaction directly undermines the accumulation and utilization of CR in later-life, weakening its protective role against cognitive decline [45]. Moreover, the high prevalence of chronic diseases, deterioration in physical functioning, and increasing mental health problems (e.g., depression) further impair cognitive functioning in the elderly [39]. The uneven distribution of healthcare resources and inadequate social security systems hinder older adults from receiving necessary support during health crises. Compounding these issues, exacerbating both physical and cognitive decline [34].
Policymakers should prioritize improving the circumstances of low-income children and families, mainly through legislation to enhance educational equity. Promoting employment equity policies and accelerating social mobility are also critical measures. Strengthening “health repair mechanisms” throughout the second half of life is essential. Establishing age-friendly public spaces or dedicated social centers for older adults and reinforcing their social networks and activities will improve CR and overall health.
The study also found that certain results from regression analyses on CR and cognitive function trajectories were significant only in continuous analyses, not in categorical ones. This may be due to the smaller differences in CR among participants and the model’s greater sensitivity to subtle changes when using continuous variables. For instance, specific midlife cognitive scores did not significantly vary with disease incidence, yet more pronounced differences were observed when participants were grouped into low, medium, and high categories. This suggests that categorical subgroups may more effectively measure disease risk, whereas continuous variables may overlook these distinctions.
It is also important to consider the potential interaction between CR and depressive states, particularly when comparing individuals with low-rapid decline cognition and those with high-stable cognition. Previous studies have shown that CR may reduce the risk of dementia associated with depression, possibly by mitigating depression-related neuroinflammation [46]. These findings suggest that individuals predisposed to depression may also experience cognitive benefits from higher levels of CR. While earlier research reported differences in the relationship between CR and the rate of cognitive decline among rural and urban older adults [47], this was not observed in the present study, potentially due to differences in the selection of CR proxies.
Limitations and future directions
One significant advantage of using latent variables in this study is the more accurate measurement of current CR-related factors at each stage, minimizing recall bias and ensuring valid CR assessments throughout the Life course. However, certain Limitations should be acknowledged. First, although CR proxies are widely accepted, they may not fully capture CR in this specific older adult population, which could limit the generalizability of these findings to other populations or contexts. Nevertheless, the use of SEM to extract CR mitigated some potential errors. Second, the small sample size of very elderly participants may have reduced the statistical power of our analyses. However, we employed more appropriate analytical techniques, such as multilevel modeling and multiple robustness checks, to mitigate potential bias. Additionally, the study had certain limitations. On one hand, the cognitive test questionnaires in the CHARLS dataset are relatively brief and may exhibit a ceiling effect, which could limit their ability to detect cognitive impairment. On the other hand, the lack of cognitively relevant biological indicators, such as neuroimaging or biomarkers, constrained our ability to assess cognitive reserve and its relationship with cognitive functioning comprehensively. Despite these limitations, the large sample size and 10-year follow-up period allowed for detailed modeling of cognitive decline over time. Furthermore, given that slowed gait speed is not only a predictor of falls, disability, sarcopenia, and mortality but also a key indicator in diagnosing dementia, we incorporated gait speed and other physical measures to enhance the reliability of our results [28].Finally, later-life CR proxy indicators (such as intellectual or social activities) may be reduced due to preclinical cognitive impairment, posing a risk of reverse causality. However, we have made every effort to exclude baseline dementia/Parkinson’s disease patients and older adults with extremely low cognitive ability to mitigate reverse causality.
Conclusion
In conclusion, this study identifies three distinct cognitive trajectories in older adults over a 10-year period, with higher CR significantly more likely to be associated with more favorable trajectories. The life-course framework confirms the cumulative and dynamic nature of CR, where early-life education, midlife health insurance, and late-life intellectual activities emerge as key proxies. These findings underscore the critical role of lifespan CR factors in maintaining cognitive function, offering valuable insights for promoting healthy aging globally through targeted interventions across life stages.
Supplementary Information
Acknowledgements
We would like to express our sincere gratitude to Mr. Binyang Tian for his valuable contribution to the data cleaning and preprocessing in this study. His meticulous work in organizing and refining the dataset greatly facilitated the subsequent analysis and ensured the reliability of our research results. Although not listed as an author, Mr. Tian’s expertise and dedication were instrumental in laying a solid foundation for the data-driven components of this project. We deeply appreciate his support and acknowledge the significance of his contributions to the completion of this work.
Abbreviations
- BMI
Body mass index
- CHARLS
China Health and Retirement Longitudinal Study
- CR
Cognitive reserve
- MCR
Motoric cognitive risk
- OR
Odds ratio
- MCI
Mild Cognitive Impairment
- AD
Alzheimer disease
- CI
Confidence Interval
- IQR
interquartile range
- MMSE
Mini-Mental State Examination
- SE
Standard error
- SEM
Structural equation modeling
- CFA
Confirmatory factor analysis
Authors’ contributions
XW, YW, and HH were responsible for the conceptualization of the study; YA and PZ conducted the data analysis; XW, GC, and YW were in charge of data interpretation. XW drafted the original manuscript, while XW, YA, PZ, and GC contributed to the review and editing. XW and GC are designated as co-first authors due to their equal contributions to the research. YW and HH are designated as co-corresponding authors for their significant roles in guidance and decision-making throughout the research process. All individuals listed as authors meet the criteria for authorship, and no others who meet these criteria have been omitted.
Funding
This paper is supported by The National Natural Science Fund (No 72374068, 82405530); Ministry of Education Industry-University Collaboration and Collaborative Education Project (No.231001282193729); Hubei Provincial Administration of Traditional Chinese Medicine Science and Technology Research Fund(No.: ZY2025L223); Science and Technology Incubation Project of the Research Institute of Traditional Chinese Medicine Big Data Knowledge Engineering, Hubei University of Chinese Medicine(No.: ZYGC-FH-202405).
Data availability
Detailed design, sampling procedures, and data collection processes for the CHARLS project can be accessed in prior publications and on the official website (http://www.icpsr.umich.edu/icpsrweb/NACDA/studies/36179), and also can be directed to the corresponding authors. Data and code will be made available on request.
Declarations
Ethics approval and consent to participate
The IRB approval number for the main household survey, including anthropometrics, is IRB00001052-11015; the IRB approval number for biomarker collection was IRB00001052-11014. All participants provided written informed consent. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments.
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.
Xiaotong Wang and Pengjun Zhou contributed equally to this work.
Contributor Information
Gao Chen, Email: chengao2046@126.com.
Hui Hu, Email: zhongyi90@163.com.
References
- 1.Chen X, Giles J, Yao Y, Yip W, Meng Q, Berkman L, et al. The path to healthy ageing in China: a Peking University–Lancet commission. Lancet. 2022;400:1967–2006. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Bhagwat N, Viviano J, Voineskos A, Chakravarty M. Alzheimer’s dis neuroimaging. Modeling and prediction of clinical symptom trajectories in alzheimer’s disease using longitudinal data. PLoS Comput Biol. 2018;14(9):e1006376. [DOI] [PMC free article] [PubMed]
- 3.Stern Y. What is cognitive reserve? Theory and research application of the reserve concept. J Int Neuropsychol Soc. 2002;8:448–60. [PubMed] [Google Scholar]
- 4.Blumen HM, Jayakody O, Ayers E, Barzilai N, Habeck C, Milman S, et al. Cognitive reserve proxies are associated with age-related cognitive decline – not age-related gait speed decline. Neurobiol Aging. 2024;141:46–54. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Bettcher BM, Gross AL, Gavett BE, Widaman KF, Fletcher E, Dowling NM, et al. Dynamic change of cognitive reserve: associations with changes in brain, cognition, and diagnosis. Neurobiol Aging. 2019;83:95–104. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Finlayson K, Miaskowski C, Alexander K, Liu W-H, Aouizerat B, Parker C, et al. Distinct wound healing and quality-of-life outcomes in subgroups of patients with venous leg ulcers with different symptom cluster experiences. J Pain Symptom Manage. 2017;53:871–9. [DOI] [PubMed] [Google Scholar]
- 7.Nyberg L, Boraxbekk C-J, Sörman DE, Hansson P, Herlitz A, Kauppi K, et al. Biological and environmental predictors of heterogeneity in neurocognitive ageing: evidence from betula and other longitudinal studies. Ageing Res Rev. 2020;64:101184. [DOI] [PubMed] [Google Scholar]
- 8.Stern Y. How can cognitive reserve promote cognitive and neurobehavioral health?? Arch Clin Neuropsychol. 2021;36:1291–5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Bartrés-Faz D, Solé-Padullés C, Junqué C, Rami L, Bosch B, Bargalló N, et al. Interactions of cognitive reserve with regional brain anatomy and brain function during a working memory task in healthy elders. Biol Psychol. 2009;80:256–9. [DOI] [PubMed] [Google Scholar]
- 10.Stern Y, Barnes CA, Grady C, Jones RN, Raz N. Brain reserve, cognitive reserve, compensation, and maintenance: operationalization, validity, and mechanisms of cognitive resilience. Neurobiol Aging. 2019;83:124–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Xu H, Yang R, Dintica C, Qi X, Song R, Bennett DA, et al. Association of lifespan cognitive reserve indicator with the risk of mild cognitive impairment and its progression to dementia. Alzheimers Dement. 2020;16:873–82. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Amanollahi M, Amanollahi S, Anjomshoa A, Dolatshahi M. Mitigating the negative impacts of aging on cognitive function; modifiable factors associated with increasing cognitive reserve. Eur J Neurosci. 2021;53(9):3109–24. [DOI] [PubMed] [Google Scholar]
- 13.Lee DH, Seo SW, Roh JH, Oh M, Oh JS, Oh SJ, et al. Effects of cognitive reserve in alzheimer’s disease and cognitively unimpaired individuals. Front Aging Neurosci. 2021;13:784054. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Song S, Stern Y, Gu Y. Modifiable lifestyle factors and cognitive reserve: a systematic review of current evidence. Ageing Res Rev. 2022;74:101551. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Corbo I, Marselli G, Di Ciero V, Casagrande M. The protective role of cognitive reserve in mild cognitive impairment: A systematic review. J Clin Med. 2023;12:1759. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Nelson ME, Jester DJ, Petkus AJ, Andel R. Cognitive, reserve. Alzheimer’s neuropathology, and risk of dementia: a systematic review and meta-analysis. Neuropsychol Rev. 2021;31:233–50. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Li X, Song R, Qi X, Xu H, Yang W, Kivipelto M, et al. Influence of cognitive reserve on cognitive trajectories: role of brain pathologies. Neurology. 2021;97:e1695–706. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Durrani R, Friedrich MG, Schulze KM, Awadalla P, Balasubramanian K, Black SE, et al. Effect of cognitive reserve on the association of vascular brain injury with cognition. Neurology. 2021;97:e1707–16. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Zhao Y, Hu Y, Smith JP, Strauss J, Yang G. Cohort profile: the China health and retirement longitudinal study (CHARLS). Int J Epidemiol. 2014;43:61–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Chen X, Wang Y, Strauss J, Zhao Y. China health and retirement longitudinal study (CHARLS). In: Gu D, Dupre ME, editors. Encyclopedia of gerontology and population aging. Cham: Springer International Publishing; 2019. pp. 1–9. [Google Scholar]
- 21.Li Y, Liu Q, Si H, Zhou W, Yu J, Bian Y, et al. Effects of (pre)frailty and cognitive reserve on mild cognitive impairment among community-dwelling older adults. Arch Gerontol Geriatr. 2024;126:105533. [DOI] [PubMed] [Google Scholar]
- 22.Xu X, Xu Y, Shi R. Association between obesity, physical activity, and cognitive decline in Chinese middle and old-aged adults: a mediation analysis. BMC Geriatr. 2024;24:54. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Yegla B, Foster TC. Operationally defining cognitive reserve genes. Neurobiol Aging. 2022;110:96–105. [DOI] [PubMed] [Google Scholar]
- 24.Camprodon-Boadas P, Prisco MD, Rabelo-da-Ponte FD, Sugranyes G, Clougher D, Baeza I, et al. Cognitive reserve and cognition in mood disorders: a systematic review and meta-analysis. Psychiatry Res. 2024;339:116083. [DOI] [PubMed] [Google Scholar]
- 25.Huoyun Z, Shilong M, Zhaoqi L, Huiqin X. Early socioeconomic status, social mobility and cognitive trajectories in later life: a life course perspective. Econ Hum Biol. 2023;50:101281. [DOI] [PubMed] [Google Scholar]
- 26.Li Y, Dekhtyar S, Grande G, Kalpouzos G, Gregorio C, Laukka EJ, et al. Association of cognitive reserve with transitions across cognitive states and death in older adults: a 15-year follow-up study. Alzheimers Dement. 2024;20:4737–46. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Dove A, Yang W, Dekhtyar S, Guo J, Wang J, Marseglia A, et al. High cognitive reserve attenuates the risk of dementia associated with cardiometabolic diseases. Alzheimers Res Ther. 2024;16(1):161. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Stark JH, Hiersche KJ, Hayes SM. 30 exploring the differential importance of modifiable fitness variables on cognitive performance in older adults. J Int Neuropsychol Soc. 2023;29:341–2. [Google Scholar]
- 29.Zhong T, Li S, Liu P, Wang Y, Chen L. The impact of education and occupation on cognitive impairment: a cross-sectional study in China. Front Aging Neurosci. 2024;16:1663–4365. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Holtzer R, Zhu X, Rosso AL, Rosano C. Cognitive reserve and risk of mobility impairment in older adults. J Am Geriatr Soc. 2022;70:3096–104. [DOI] [PubMed] [Google Scholar]
- 31.Liu I, Huang Y-J, Wang L-K, Tsai Y-H, Hsu S-L, Chang C-J, et al. Dual trajectories of loneliness and depression and their baseline correlates over a 14-year follow-up period in older adults: results from a nationally representative sample in Taiwan. Int J Older People Nurs. 2021;16:e12410. [DOI] [PubMed] [Google Scholar]
- 32.Nagin DS, Jones BL, Passos VL, Tremblay RE. Group-based multi-trajectory modeling. Stat Methods Med Res. 2018;27:2015–23. [DOI] [PubMed] [Google Scholar]
- 33.Nagin DS, Odgers CL. Group-Based trajectory modeling in clinical research. Ann Rev Clin Psychol. 2010;6:109–38. [DOI] [PubMed] [Google Scholar]
- 34.Zhou S, Song S, Jin Y, Zheng Z-J. Prospective association between social engagement and cognitive impairment among middle-aged and older adults: evidence from the China health and retirement longitudinal study. BMJ Open. 2020;10:e040936. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Yuan Y, Lapane KL, Tjia J, Baek J, Liu S-H, Ulbricht CM. Trajectories of physical frailty and cognitive impairment in older adults in United States nursing homes. BMC Geriatr. 2022;22:339. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Rosenberg DE, Wu Y, Idu A, Greenwood-Hickman MA, McCurry SM, LaCroix AZ, et al. Historic cognitive function trajectories as predictors of sedentary behavior and physical activity in older adults. Journals Gerontology: Ser A. 2024;15:glae125. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Ye P, Jin Y, Er Y, Duan L, Palagyi A, Fang L, et al. A scoping review of National policies for healthy ageing in Mainland China from 2016 to 2020. Lancet Reg Health. 2021;12:100168. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Greenfield PM, Brown G, Du H. Shifts in ecology, behavior, values, and relationships during the coronavirus pandemic: survival threat, subsistence activities, conservation of resources, and interdependent families. Curr Res Ecol Soc Psychol. 2021;2:100017. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Livingston G, Huntley J, Liu KY, Costafreda SG, Selbæk G, Alladi S, et al. Dementia prevention, intervention, and care: 2024 report of the lancet standing commission. Lancet. 2024;404:572–628. [DOI] [PubMed] [Google Scholar]
- 40.Jia L, Du Y, Chu L, Zhang Z, Li F, Lyu D, et al. Prevalence, risk factors, and management of dementia and mild cognitive impairment in adults aged 60 years or older in china: a cross-sectional study. Lancet Public Health. 2020;5:e661–71. [DOI] [PubMed] [Google Scholar]
- 41.Skoblow HF, Proulx CM, Akpolat R, Palermo F. Early-life socioeconomic position and later-life cognitive functioning: a meta-analysis. Soc Sci Med. 2024;359:117267. [DOI] [PubMed] [Google Scholar]
- 42.Magdaleno Herrero R, Murillo-García N, Yorca-Ruiz Á, Neergaard K, Crespo-Facorro B, Ayesa-Arriola R. Biomarkers as proxies for cognitive reserve: the role of high density lipoprotein cholesterol in first episode of psychosis. Span J Psychiatry Ment Health. 2024;17:146–53. [DOI] [PubMed] [Google Scholar]
- 43.Yan Y, Du Y, Li X, Ping W, Chang Y. Physical function, ADL, and depressive symptoms in Chinese elderly: evidence from the CHARLS. Front Public Health. 2023;11:1017689. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Zhang W, Wang T, Wang A. Impact of physical activity intensity on longitudinal trajectories of cognitive function and depressive symptoms in middle-aged and older Chinese adults: eight-year prospective study. J Affect Disord. 2022;315:64–9. [DOI] [PubMed] [Google Scholar]
- 45.Xiang X, Lai PHL, Bao L, Sun Y, Chen J, Dunkle RE, et al. Dual trajectories of social isolation and dementia in older adults: a population-based longitudinal study. J Aging Health. 2021;33:63–74. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46.Xu X, Zheng Y, Fang J, Huang J, Yang X, Zhu X, et al. Associations between regular physical exercise and physical, emotional, and cognitive health of older adults in China: an 8-year longitudinal study with propensity score matching. Front Public Health. 2024;12:1301067. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47.Chen X, Xue B, Hu Y. Cognitive reserve over life course and 7-year trajectories of cognitive decline: results from China health and retirement longitudinal study. BMC Public Health. 2022;22:231. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
Detailed design, sampling procedures, and data collection processes for the CHARLS project can be accessed in prior publications and on the official website (http://www.icpsr.umich.edu/icpsrweb/NACDA/studies/36179), and also can be directed to the corresponding authors. Data and code will be made available on request.