Abstract
Background
Physical activity (PA) has been associated with depressive symptoms (DS), yet most evidence relies on single time-point assessments and general populations. It remains unclear whether long-term trajectories of PA are associated with the odds of DS among older adults with chronic disease (participants with at least one chronic disease), including those with multimorbidity (participants with two or more chronic diseases). Therefore, this study aimed to examine the association between trajectories of PA and DS with chronic disease.
Methods
A total of 854 older adults (≥ 65 years) with chronic disease (participants with at least one chronic disease) derived from the Chinese Longitudinal Healthy Longevity Survey (CLHLS) in three waves 2011–2012, 2014, and 2017–2018. PA trajectories were identified using group-based trajectory modeling across three waves. Participants who were free of DS at baseline were followed up at the third wave to assess DS status. Multivariable-adjusted binary logistic regression models were applied to estimate odds ratios (ORs) and 95% confidence intervals (CIs), adjusting for demographic, socioeconomic, lifestyle, and health-related covariates. Subgroup analyses were conducted among individuals with multimorbidity and specific chronic diseases.
Results
Three distinct PA trajectories were identified: low level and decreasing, moderate level and decreasing, and high level and decreasing. Compared with the low level and decreasing trajectory, both moderate level and decreasing and high level and decreasing trajectories were associated with lower odds of DS. These associations were generally consistent among individuals with multimorbidity. For specific chronic diseases, statistically significant associations were observed in participants with hypertension, lung disease, stroke, and arthritis, while similar trends were noted but did not reach statistical significance in other conditions.
Conclusions
Maintaining favorable long-term patterns of PA, even with gradual decline, is associated with reduced odds of DS among older adults with chronic diseases. These findings support PA as a potentially cost-effective strategy for secondary prevention and highlight the importance of sustained behavioral patterns in mitigating mental health risks in aging populations.
Supplementary Information
The online version contains supplementary material available at 10.1186/s12877-026-07797-w.
Keywords: Physical activity, Depressive symptoms, Chronic disease, Physical-psychological comorbidity, Older adults, CLHLS
Introduction
With the rapid aging process of the worldwide population, the burden of depressive symptoms (DS) in the older adults on individuals, families, and society is increasing, and high detection of DS exaggerate the loss of healthy life expectancy [1]. Existing evidence indicates DS often typically lead to somatic symptoms of emotional distress, leading to multisystem comorbidity in individuals [2], such as co-occurring physical, psychological, and cognitive multimorbidity [3, 4], and further contributing to functional disability [5]. Therefore, it is essential that effective and modifiable prevention strategies of DS are targeted for older adults.
Physical activity (PA) is defined as nay movement of the body produced by skeletal muscles that results in a measurable increase in energy consumption [6]. Increasing evidence found that individuals who keep PA have a decreased risk of DS. Evidence from Schuch et al. [7] suggested a significant inverse relationship between PA and depression, specifically, a meta-analysis of forty-nine prospective studies found that high PA levels correlate with a 17% lower risk of depression. This conclusion is reinforced by various other meta-analytic reviews that identify physical inactivity as a notable risk factor for the condition [8, 9]. And, small amounts of fragmented PA have still proven effective for the condition [10]. However, existing literature yielded inconsistent findings regarding this correlation; consequently, the link between PA and DS has yet to be firmly established [11, 12]. Moreover, although most prior research found how baseline PA associated with DS [13–18], the role of PA change over time is still unclear. It is as PA always present the dynamic characteristics over time [19–21], and reduces concern about residual confounding for PA methodologically. It is more reasonable to explore the role on DS through identifying the long-term trajectories of PA.
Exploring PA-based intervention strategies to reduce DS is particularly important among older adults with chronic diseases. Compared with the general older population, individuals with chronic diseases are more vulnerable to DS due to accumulated symptom burden, functional limitations, reduced mobility, long-term treatment demands, and health-related stressors. This issue may become even more pronounced among individuals with multimorbidity, where the coexistence of multiple chronic conditions substantially increases physical, psychological, cognitive multimorbidity, and social burden [3, 4]. Moreover, some chronic diseases, particularly those associated with mobility limitation, pain burden, dyspnea, or disability-related symptoms, may make sustained engagement in PA both more challenging and more clinically relevant for psychological well-being. In this context, DS may not only impair quality of life, but also worsen self-management behaviors, treatment adherence, and disease progression, thereby contributing to more complex physical-psychological multimorbidity states [2].
Within current chronic disease management frameworks, secondary prevention aims not only to delay disease progression and reduce complications, but also to preserve psychological well-being and functional capacity. PA has been recognized as a potentially accessible and cost-effective strategy that may provide simultaneous physical and psychological benefits in populations with chronic diseases [22]. However, existing studies have primarily focused on the association between baseline PA and DS in the general population [13–18], leaving uncertainty regarding whether long-term patterns of PA are similarly associated with DS among older adults with chronic condition, including those with multimorbidity and specific chronic conditions.
This evidence gap is particularly relevant in China, where rapid population aging and increasing multimorbidity continue to place substantial pressure on healthcare system and long-term care resource. As chronic diseases progress and accumulate, healthcare needs become increasingly complex and costly. Therefore, identifying accessible and cost-effective behaviors strategies, such as maintaining favorable PA patterns, may help reduce the psychological burden associated with chronic disease and support healthy aging in resource-constrained settings.
To address these research gaps, we used data from a nationally representative survey in China to examine the association between long-term trajectories of PA over 7-year and DS among older adults with chronic diseases (participants with at least one chronic disease). This investigation also responds to ongoing national health initiatives, such as “Healthy China 2030”, which emphasize PA promotion and chronic disease management as key components of healthy aging strategies. By evaluating longitudinal patterns of PA rather than relying solely on single baseline measurements, our study aims to provide more robust evidence to inform targeted and cost-effective intervention strategies for older adults with chronic diseases.
Methods
Study design and participants
This longitudinal study was conducted within the Chinese Longitudinal Healthy Longevity Survey (CLHLS), which is an ongoing nationally representative population-based survey and covered the countries and cities in 23 provinces in Chinese older adults. Details of study design and sampling have been well-documented in prior research [23]. Briefly, the baseline of CLHLS was performed in 1998 and followed up and further recruitments of new participants in 2000, 2002, 2005, 2008–2009, 2011–2012, 2014, and 2017–2018 [24]. For this current study, we used the 3 waves of CLHLS 2011–2018 data. These waves were selected because comparable measurements of PA were consistently available across the 2011–2012, 2014, and 2017–2018 surveys, enabling the identification of longitudinal PA trajectories over time.
Several exclusion criteria were established: (1) Age < 65 years (baseline); (2) missing formation on PA during CLHLS 2011–2018; (3) participants without chronic diseases (CLHLS 2011–2012, baseline); (4) participants with DS (CLHLS 2011–2012, baseline), and (5) missing formation on DS in CLHLS 2017–2018 (Fig. 1).
Fig. 1.
Study design. A Timeline of the analysis. B Flowchart of participants included in this study
In this present study, we included seven types of self-reported physician-diagnosed major chronic diseases from these abovementioned surveys: hypertension, diabetes, cancer, lung disease, heart disease (including heart attack, coronary heart disease, angina, congestive heart failure, or other heart problems), stroke, and arthritis [25–27]. If participants reported 2 or more chronic disease, they are classified as having multimorbidity. Because only 10 participants reported a physician diagnosis of cancer at baseline, cancer-specific analyses were not performed due to insufficient statistical power for reliable estimation.
Exposure
Trajectories of PA were the exposure of interest. The definition of PA follows previous research [24, 28]. PA is assessed at same questionnaire and criteria in CLHLS 2011–2012, 2014, and 2017–2018, respectively.
Specifically, PA was assessed through a combination of nine activity frequencies: regular exercise, housework tasks, personal outdoor activities, gardening, reading, rearing domestic animals/pets, playing cards/mah-jongg, watching television/listening to the radio, and attending social activities. According to the frequency of various physical activities, this study calculated the total PA score, ranging from 0 to 18. The higher PA score presents higher level PA. Details of PA score are provided in Table S1.
Outcome
The outcome of interest is DS. DS is measured at different scales in in CLHLS 2011–2012 and 2017–2018, separately.
For CLHLS 2011–2012, DS was measured through the Phenotypes and eXposures (PhenX) toolkit, which was developed by integrating criteria from the Diagnostic and Statistical Manual of Mental Disorders (5th edition; DSM-5) and the Comprehensive International Diagnostic Interview Screening Scale (CIDI-SC) [29, 30]. It utilizes two concise binary (yes or no) self-report items to evaluate participants’ experiences in the past year: (1) persistent feelings of sadness or depression lasting at least 14 days, and (2) a diminished interest in routine activities or hobbies. The presence of DS is identified if a respondent provides a positive response (yes) to either question.
For CLHLS 2017–2018, DS was assessed through the 10-item Center for Epidemiological Studies Depression Scale (CESD-10), which have been widely applied in Chinese older adults [31] and has excellent validity and reliability in CLHLS [32]. The CESD-10 scale includes 10 items, including 8 negative items were coded as 0 (< 1 day), 1 (1–2 days), 2 (3–4 days), and 3 (5–7 days), while two item (feeling hopeful and happy) were reverse-coded. Based on these calculation criteria, the total score of CESD-10 ranges from 0 to 30, a higher score presents more severe symptoms. According to previous research, the presence of DS is defined as the total score ≥ 10 [33].
Although distinct measurement tools were used to measure DS at baseline and endpoint, the PhenX and the CESD-10 have excellent effectiveness and reliability, and this approach has been widely used in previous studies [34, 35]. Detailed information on the PhenX and the CESD-10 are provided in Table S2.
Covariates
Covariates were assessed at the baseline in this study, including age (continuous), sex (male, female), place of residence (urban, rural), marital status (in marriage, not in marriage), education attainment (< 1 year, 1–6 years, and > 6 years), sufficient financial support (yes, no), occupation before age 60 (professional work, others), coresidence (alone, not alone), diet (healthy, unhealthy), smoking (current non-smoking, others), drinking (non-harmful alcohol consumption, others), body mass index (< 18.5, 18.5–23.9, 24.0-27.9, and ≥ 28.0), and number of chronic diseases (1, ≥ 2).
According to previous research, this study used simplified healthy eating (SHE) index to assess the diet [24, 36]. The SHE includes 9 common food types: fresh fruits, fresh vegetables, legumes, garlic, tea, meat, fish, eggs, and preserved vegetables. The total score of SHE index was calculated through the frequency of consumption of each food types. The total score ranges from 0 to 18, higher score indicates more healthy diet. The distribution of total score of SHE index in the top two-fifth as the healthy diet, otherwise unhealthy diet [28]. Body mass index was calculated through weight (kg) divided by height (m2). Details of covariate definitions were provided in Table S3 and Table S4.
Statistical analyses
In this current study, trajectories of PA were mapped from the CHLHS 2011–2018 using Group-based trajectory modeling (GBTM) to capture population heterogeneity in PA trend. By treating the data collection years as the time axis and assuming a censored normal distribution for the continuous PA measures, we identified latent subgroups with distinct development paths of PA over time. The analysis was performed in STATA, leveraging the traj package for model fitting and trajplot for visual representation [37, 38].
The optimal number of latent trajectory groups was determined through an interactive process, testing configurations from one to four classes. According to prior studies [39, 40], several selections of the final model and its associated functional forms was guided by multi-dimensional criteria: (1) minimizing the Bayesian Information Criterion (BIC), aiming for absolute values closest to zero; (2) ensuring a high degree of classification accuracy with an Average Posterior Probability (AvePP) exceeding 70%; and (3) confirming the theoretical and real plausibility of the identified patterns.
The participants’ characteristics were grouped by trajectories of PA and presented as frequencies (proportion) for categorical variables and as the median (interquartile range) for continuous variable. The chi-square tests and analysis of variance (ANONA) was used to assess the differences between trajectory groups, separately for categorical and continuous variables.
Multivariable-adjusted binary logistic regression models were used to estimate the odds ratios (ORs) and 95% confidence intervals (CIs) for the association of trajectories of PA with DS. Two models were considered in this study: (1) Model 1 is the crude model, including the exposure of interest. (2) Model 2 is the adjusted model, additionally adjusting age, sex, place of residence, marital status, education attainment, sufficient financial support, occupation before age 60, living alone, diet, smoking, drinking, body mass index, and number of chronic diseases. The interpretation of the results is based on model 2. These models were not only performed in the whole study sample with chronic disease (regardless of the number of chronic diseases), but also among older adults with multimorbidity, to test whether the protective role of PA for DS in the context of higher burden of chronic disease is still valid. Moreover, we also conducted same analyses among older adults with specific chronic disease (including hypertension, diabetes, cancer, lung disease, heart disease, stroke, and arthritis).
Moreover, sensitivity analyses based on complete-case datasets were performed to examine the robustness of the findings with respect to the imputation approach. Missing values in covariates were addressed using multiple imputation by chained equations (MICE) under the missing-at-random assumption, with five imputed datasets generated. Estimates were subsequently combined using Rubin’s rules to obtain overall effect estimates [41].
Statistical analyses were performed using Stata, version MP18 (StataCorp LLC). Statistical significance is P < 0.05 (2-tailed).
Results
Systematic evaluation of the model selection parameters led to the adoption of a 3-group solution, with each trajectory modeled using a second-order polynomial shape (2 2 2). This configuration achieved the superior balance between statistical rigor and substantive interpretability. The underlying fit statistics supporting this selection are provided in Table S5. Three trajectories of PA were identified and characterized by “Low level and decreasing” (n = 106, 12.41%), “Moderate level and decreasing” (n = 551, 64.52%), and “High level and decreasing” (n = 197, 23.07%) (Fig. 2). Among these trajectories of PA, they have similarity in decreasing trend over time, but the difference is having distinct PA level at baseline.
Fig. 2.
Trajectories of physical activity from 2011–2018. The solid lines display the estimated values and the dotted lines represent the 95% confidence intervals. Low-decline = Low level and decreasing; Moderate-decline=Moderate level and decreasing; High-decline=High level and decreasing
Table 1 shows the group-difference in trajectories of PA in characteristics of participants with chronic disease. Compared to participants experiencing the high level and decreasing trajectory of PA, those experienced low level and decreasing trajectory of PA were more likely to be having DS at endpoint, older, female, rural residence, not in marriage, lower education attainment, other work (not professional work) before 60, living alone, unhealthy diet, current non-smoking, non-harmful alcohol consumption.
Table 1.
Characteristics of participants with chronic disease according to trajectories of physical activity
| Characteristics | Trajectories of physical activity | P-value | ||
|---|---|---|---|---|
| Low level and decreasing (N = 106) | Moderate level and decreasing (N = 551) | High level and decreasing (N = 197) | ||
| Had depressive symptoms at endpoint | 45 (42.5) | 152 (27.6) | 27 (13.7) | < 0.001 |
| Age, median (IQR), years | 83.0 (10.0) | 76.0 (10.0) | 73.0 (6.0) | < 0.001 |
| Male | 37 (34.9) | 261 (47.4) | 127 (64.5) | < 0.001 |
| Rural residence | 70 (66.0) | 324 (58.8) | 76 (38.6) | < 0.001 |
| In marriage | 33 (31.1) | 330 (60.0) | 147 (74.6) | < 0.001 |
| Education attainment | < 0.001 | |||
| < 1 year | 77 (72.6) | 273 (49.6) | 30 (15.2) | |
| 1–6 years | 26 (24.5) | 216 (39.3) | 81 (41.1) | |
| > 6 years | 3 (2.8) | 61 (11.1) | 86 (43.7) | |
| Had sufficient financial support | 84 (79.3) | 443 (80.7) | 171 (86.8) | 0.120 |
| Professional work before 60 | 1 (1.1) | 28 (5.9) | 48 (27.4) | < 0.001 |
| Living alone | 32 (30.5) | 112 (20.5) | 20 (10.2) | < 0.001 |
| Healthy diet | 22 (20.8) | 175 (31.8) | 105 (53.3) | < 0.001 |
| Current non-smoking | 76 (71.7) | 359 (65.2) | 104 (52.8) | 0.001 |
| Non-harmful alcohol consumption | 96 (90.6) | 480 (87.1) | 154 (78.2) | 0.003 |
| Body mass index | 0.225 | |||
| < 18.5 | 19 (18.3) | 71 (13.0) | 17 (8.7) | |
| 18.5–23.9 | 54 (51.9) | 284 (51.8) | 97 (49.7) | |
| 24.0-27.9 | 23 (22.1) | 139 (25.4) | 57 (29.2) | |
| ≥ 28.0 | 8 (7.7) | 54 (9.9) | 24 (12.3) | |
| Number of chronic diseases (≥ 2) | 31 (29.3) | 187 (33.9) | 77 (39.1) | 0.202 |
Missing information: 1 for education attainment, 1 for marital status, 2 for sufficient financial support, 114 for occupation before age 60, 7 for living alone, 7 for body mass index
Data are n (%), unless otherwise indicated
Abbreviation: IQR Interquartile range
Table 2 presents the longitudinal association between trajectories of PA and DS among older adults with chronic disease. For the analysis among older adults with chronic disease (including all participants with at least one chronic disease), participants with both moderate level and decreasing and high level and decreasing trajectories had lower odds of experiencing DS, than those with the low level and decreasing trajectory, especially for the high level and decreasing trajectory of PA. Specifically, there were 52% lower odds of DS in the moderate level and decreasing trajectory (OR = 0.48, 95%CI = 0.29–0.80) and 80% lower odds of DS in the high level and decreasing trajectory (OR = 0.20, 95%CI = 0.10–0.41). Meanwhile, we further explored these association in participants with multimorbidity (participants with two or more chronic diseases), and then the similar findings were found. Compared to participants with the low level and decreasing trajectory, those with both moderate level and decreasing (OR = 0.17, 95%CI = 0.06–0.47) and high level and decreasing (OR = 0.12, 95%CI = 0.04–0.41) had lower odds of experiencing DS (Table 2). In the sensitivity analysis of multiple imputation, similar findings were found (Table S6).
Table 2.
Association between trajectories of physical activity and depressive symptoms among older adults with chronic disease
| Trajectories of physical activity | Odds Ratio (95%CI) | |
|---|---|---|
| Model 1 | Model 2 | |
| For the analysis among older adults with chronic disease (including all participants with at least one chronic disease) (n = 854) | ||
| Low level and decreasing | 1 [Reference] | 1 [Reference] |
| Moderate level and decreasing | 0.52 (0.34–0.79) ** | 0.48 (0.29–0.80) ** |
| High level and decreasing | 0.22 (0.12–0.38) *** | 0.20 (0.10–0.41) *** |
| For the analysis among older adults with chronic diseases comorbidity (participants with two or more chronic diseases) (n = 295) | ||
| Low level and decreasing | 1 [Reference] | 1 [Reference] |
| Moderate level and decreasing | 0.28 (0.13–0.62) ** | 0.17 (0.06–0.47) *** |
| High level and decreasing | 0.18 (0.07–0.46) *** | 0.12 (0.04–0.41) *** |
Model 1 was crude model
Model 2 was adjusted for age, sex, place of residence, marital status, education attainment, sufficient financial support, occupation before age 60, living alone, diet, smoking, drinking, body mass index, and number of chronic diseases
For model 2, for the analysis of among older adults with chronic disease ((including all participants with at least one chronic disease), the number of chronic diseases was categorical variable (1, ≥2); while for the analysis of among older adults with chronic disease comorbidity (participants with two or more chronic diseases), the number of chronic diseases was continuous variable
**P<0.01, ***P<0.001
Moreover, we found that the protection role of healthier trajectories of PA for the DS odds in participants with most specific chronic disease still existed controlling for age, sex, place of residence, marital status, education attainment, sufficient financial support, occupation before age 60, living alone, diet, smoking, drinking, body mass index, and number of chronic diseases (continuous). Compared to those with hypertension and with the low level and decreasing trajectory of PA, individuals with hypertension in the high level and decreasing trajectory (OR = 0.26, 95%CI = 0.10–0.66) experienced significantly lower odds of DS, while those with the high level and decreasing trajectory of PA and lung diseases (OR = 0.14, 95%CI = 0.03–0.62), stroke (OR = 0.04, 95%CI = 0.01–0.79), arthritis (OR = 0.08, 95%CI = 0.02–0.37) had markedly decreased odds of DS. For moderate level and decreasing trajectories, the odds of DS was pronounced in individuals with lung diseases, stroke, and arthritis (Table 3).
Table 3.
Association between trajectories of physical activity and depressive symptoms among older adults with specific chronic disease
| Trajectories of physical activity | Odds Ratio (95%CI) | |
|---|---|---|
| Model 1 | Model 2 | |
| Hypertension (n = 506) | ||
| Low level and decreasing | 1 [Reference] | 1 [Reference] |
| Moderate level and decreasing | 0.60 (0.34–1.07) | 0.51 (0.26–1.03) |
| High level and decreasing | 0.27 (0.13–0.57) *** | 0.26 (0.10–0.66) ** |
| Diabetes (n = 69) | ||
| Low level and decreasing | 1 [Reference] | 1 [Reference] |
| Moderate level and decreasing | 0.59 (0.08–4.63) | 0.01 (0.00-1.43) |
| High level and decreasing | 0.16 (0.02–1.59) | 0.01 (0.00-3.34) |
| Lung diseases (n = 168) | ||
| Low level and decreasing | 1 [Reference] | 1 [Reference] |
| Moderate level and decreasing | 0.38 (0.15–0.97) * | 0.31 (0.10–0.94) * |
| High level and decreasing | 0.19 (0.06–0.65) ** | 0.14 (0.03–0.62) ** |
| Heart diseases (n = 173) | ||
| Low level and decreasing | 1 [Reference] | 1 [Reference] |
| Moderate level and decreasing | 0.33 (0.11–0.96) * | 0.40 (0.10–1.61) |
| High level and decreasing | 0.18 (0.05–0.63) ** | 0.22 (0.04–1.15) |
| Stroke (n = 106) | ||
| Low level and decreasing | 1 [Reference] | 1 [Reference] |
| Moderate level and decreasing | 0.57 (0.17–1.89) | 0.12 (0.01–0.98) * |
| High level and decreasing | 0.25 (0.05–1.14) | 0.04 (0.01–0.79) * |
| Arthritis (n = 234) | ||
| Low level and decreasing | 1 [Reference] | 1 [Reference] |
| Moderate level and decreasing | 0.32 (0.14–0.73) ** | 0.15 (0.05–0.43) *** |
| High level and decreasing | 0.18 (0.06–0.52) *** | 0.08 (0.02–0.37) *** |
Model 1 was crude model
Model 2 was adjusted for age, sex, place of residence, marital status, education attainment, sufficient financial support, occupation before age 60, living alone, diet, smoking, drinking, body mass index, and number of chronic diseases (continuous)
*P < 0.05, **P < 0.01, ***P < 0.001
Abbreviation: CI Confidence interval
Cancer-specific analyses were not conducted because only 10 participants reported cancer at baseline, resulting in insufficient statistical power for reliable estimation
Discussion
In this study, three different long-term trajectories of PA were identified. Older Chinese adults with chronic disease (regardless of number) and healthier trajectories (moderate level and decreasing, and high level and decreasing) of PA were less likely to experience DS than low level and decreasing PA trajectory. Similar findings were also found in the older adults with multimorbidity and several specific chronic diseases. The main findings demonstrate that maintaining relatively favorable PA may be important for reducing the odds of DS among older adults with chronic diseases.
Substantial epidemiological evidence has suggested the association between PA at single time point and DS. Short-term (3 months) observational study among 755 adults in the United States indicated that moderate-vigorous PA at baseline significantly correlated with lower DS (r=-0.19) [13], a finding mirrored in multi-year cohort studies across both Western and Asian populations. A 13-year cohort study involving 10,325 adults in Norway found that light (β = 0.11) or moderate-vigorous PA (β = 0.19) played a protective role against psychological problem during follow-up [16], while another 32-year cohort study among 676 women in Gothenburg, Sweden, showed that lower baseline PA levels were associated with higher depression scores [18]. Recent prospective cohort studies conducted in South Korea and China have also found significant associations between high PA levels and lower odds of DS [15, 17]. However, this association has not been universally corroborated, as some observational studies found that the predictive effect of baseline PA diminished over time [11, 12]. These discrepancies may partly reflect the limitations of relying on single baseline assessments of PA, which cannot account for long-term behavioral patterns. By incorporating repeated PA assessments over three survey waves, the present study extends previous evidence by examining sustained PA patterns during older adulthood rather than relying solely on static baseline measures. However, the identified trajectory groups mainly differed in overall PA level rather than in the rate of decline, as all groups showed generally decreasing patterns over time. Therefore, the observed associations may primarily reflect the long-term maintenance of relatively higher PA levels during older adulthood, rather than the independent effect of decline slope. This trajectory-based approach provides additional insight into the association between sustained PA patterns and DS in later life [19–21]. Moreover, given that the potential bidirectional relationships between PA and DS, we followed-up the DS-free population at baseline to strengthen the temporal sequence between trajectories of PA and subsequent DS [31].
Focusing on populations with chronic diseases, particularly those with multimorbidity, is clinically important because these populations often experience greater symptom burden, reduced mobility, functional limitation, and psychological vulnerability. In this context, DS may contribute to poorer self-management, reduced treatment adherence, and worsening health status. Our findings therefore support the potential role of PA as a practical strategy within secondary prevention frameworks for chronic disease management. These findings are also consistent with current public health priorities, including the World Health Organization recommendations and the “Healthy China 2030” initiative, both of which emphasize PA promotion and healthy aging. Importantly, our findings suggest that maintaining achievable and sustainable PA patterns during older adulthood, rather than pursuing high-intensity activity levels, may still provide psychological benefits among individuals living with chronic diseases [42–44].
When examining specific chronic diseases, some heterogeneity in associations was observed. It should be noted that these disease-specific subgroup analyses were not mutually exclusive; therefore, participants with multimorbidity could be included in more than one chronic disease subgroup. In this study, we included seven major physician-diagnosed chronic conditions, including hypertension, diabetes, cancer, lung disease, heart disease, stroke, and arthritis. While the overall direction of association between favorable PA trajectories and lower odds of DS was generally consistent across chronic disease groups, statistically significant associations were primarily observed among individuals with hypertension, lung disease, stroke, and arthritis. Several disease-specific mechanisms may explain these findings. For individuals with hypertension, PA contributes to improved vascular health, better blood pressure control, and reduced systemic inflammation. Regarding respiratory health, PA among those with lung disease may improve exercise tolerance and reduce further adverse health outcome associated with respiratory symptoms. For those managing stroke, PA can facilitate functional recovery, enhance neuroplasticity, and promote social participation, thereby reducing vulnerability to DS. Finally, among individuals with arthritis, PA has been shown to alleviate pain, improve mobility, and strengthen self-efficacy, all of which are closely linked to enhanced psychological well-being [2, 45, 46]. Beyond these biological mechanisms, the larger sample size of these specific subgroups likely provided greater statistical power to detect these associations. In contrast, no statistically significant associations were observed among individuals with diabetes and heart disease after full adjustment. However, these findings should be interpreted cautiously. The diabetes subgroup was relatively small (n = 69), resulting in wide confidence intervals and limited statistical power. For heart disease, the associations detected in crude analyses were attenuated after multivariable adjustment. This attenuation suggests that demographic, socioeconomic, and health-related factors may partly account for the initial relationship between PA trajectories and DS. Therefore, the absence of statistical significance in these chronic disease groups should not be interpreted as definitive evidence of a lack of association. Furthermore, the reliability and validity of these findings, particularly within the diabetes subgroup, should be evaluated in future studies with larger sample sizes.
Taken together, these findings suggest that maintaining relatively favorable and achievable PA patterns during older adulthood may be associated with lower odds of DS among individuals with chronic diseases. From a public health perspective, promoting sustainable PA engagement may represent a practical strategy to support psychological well-being in aging populations with increasing multimorbidity burden.
Strengths and limitations
This study has several notable strengths. First, leveraging a large, nationally representative cohort from China enhances the generalizability of our findings to older adults in real-world settings. Second, by adopting a trajectory-based approach to characterize long-term patterns of PA over a 7-year period, this study moves beyond conventional single time-point assessments and captures the dynamic nature of behavioral exposure, thereby providing a more nuanced understanding of its relationship with the odds of DS. Third, the focus on individuals with chronic diseases, including those with multimorbidity, addresses an important yet underexplored high-risk population and aligns closely with current priorities in secondary prevention. In addition, the consistency of findings across multiple subgroups, including several specific chronic conditions, further supports the robustness of the observed associations.
Several limitations should also be acknowledged. First, PA, DS and chronic diseases were assessed based on self-reported measures, which may be subject to recall bias and misclassification. Second, although we accounted for a range of potential confounders and restricted analyses to participants free of DS at baseline to strengthen temporal inference, residual confounding and reverse causation cannot be entirely excluded due to the observational nature of the study. Third, despite the discrepancy in depression scales employed during the 2011–2012 and 2017–2018 waves, the empirical consistency of these measures has been rigorously validated for CLHLS [34, 35]. Fourth, information on the severity and duration of chronic diseases was not available, which may have contributed to heterogeneity in the observed associations across disease groups. Fifth, while trajectory modeling provides valuable insights into long-term patterns, it may still mask within-group variability and does not fully capture short-term fluctuations in PA behavior. Sixth, PA trajectories were evaluated during older adulthood only. Therefore, the identified trajectory patterns may not capture the potential influence of PA exposure during earlier stages of life, such as midlife, which may also contribute to the development of DS in later life. Seventh, covariates were assessed at baseline only, although several socioeconomic, lifestyle, and health-related characteristics may change over time in older adults. Future studies incorporating time-varying covariates are warranted to further clarify these longitudinal associations. Finally, the generalizability of these findings to non-Chinese populations or younger age groups requires further validation.
Conclusions
In conclusion, our findings highlight the importance of considering long-term patterns of PA, rather than single time-point assessments, in understanding the odds of DS among older adults with chronic diseases. Maintaining relatively favorable long-term patterns of PA during older adulthood was associated with lower odds of DS, including among individuals with multimorbidity and several specific chronic diseases. These findings support the importance of sustained and achievable PA engagement as a practical strategy for chronic disease management and healthy aging.
Supplementary Information
Supplementary Material 1. Table S1. Scoring standard of physical activity. Table S2. Scales and scoring standard of depressive symptoms. Table S3. Definition of covariates. Table S4. Definition of smoking, drinking, and diet. Table S5. Goodness-of-fit statistics of group-based trajectory analysis. Table S6. Sensitivity analysis of multiple imputation of association between trajectories of physical activity and depressive symptoms among older adults with chronic disease.
Acknowledgements
The author thanks the participants and enumerators of CLHLS for their data, and to everyone who cooperated in the survey implementation.
Declaration of generative AI
During the preparation of this work, the authors used ChatGPT-5.0 to check grammar and revise certain sentences, aiming to improve readability and language. After using this tool, the author reviewed and edited the content as needed and takes full responsibility for the content of the publication.
Abbreviations
- PA
Physical activity
- DS
Depressive symptoms
- CLHLS
The Chinese Longitudinal Healthy Longevity Survey
- ORs
Odds ratios
- CIs
Confidence intervals
- STROBE
The Strengthening the Reporting of Observational Studies in Epidemiology
- PhenX
EXposures
- DSM-5
The Diagnostic and Statistical Manual of Mental Disorders
- CIDI-SC
The Comprehensive International Diagnostic Interview Screening Scale
- CESD-10
The 10-item Center for Epidemiological Studies Depression Scale
- SHE
Simplified healthy eating
- GBTM
Group-based trajectory modeling
- BIC
Bayesian Information Criterion
- AvePP
Average Posterior Probability
- ANONA
Analysis of variance
- MICE
Multiple imputation by chained equations
Authors’ contributions
Concept and design: JH and SCP. Acquisition, analysis, or interpretation of data: JH and SCP. Drafting of the manuscript: JH and SCP. Critical review of the manuscript for important intellectual content: JH, XL, SCP. Statistical analysis: JH. Validation: JH, XL, SCP. Supervision: SCP. JH and SCP are responsible for ensuring the integrity and accuracy of the study. All authors approved the submission.
Funding
The authors received no financial support for the research, authorship, and/or publication of this work.
Data availability
All datasets used in this analysis are available for download at (https://opendata.pku.edu.cn/dataverse/CHADS).
Declarations
Ethics approval and consent to participate
This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline and the Declaration of Helsinki. The CLHLS received approval from the Ethics Committee of Peking University (IRB00001052-13074) and all participants signed the written consent. This study’s ethics clearance was obtained from the Institutional Review Board of the Faculty of Social Sciences and Humanities, Mahidol University with certificate number 2026/051.
Consent for publication
None.
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.
References
- 1.Herrman H, Patel V, Kieling C, Berk M, Buchweitz C, Cuijpers P, Furukawa TA, Kessler RC, Kohrt BA, Maj M, et al. Time for united action on depression: a Lancet-World Psychiatric Association Commission. Lancet. 2022;399(10328):957–1022. [DOI] [PubMed] [Google Scholar]
- 2.Zhou YG, Dai XC, Ni YJ, Zeng QY, Cheng YY, Carrillo-Larco RM, Yan LJL, Xu XL. Interventions and management on multimorbidity: An overview of systematic reviews. Ageing Res Rev. 2023;87:101901. [DOI] [PubMed]
- 3.Ronaldson A, de la Torre JA, Prina M, Armstrong D, Das-Munshi J, Hatch S, Stewart R, Hotopf M, Dregan A. Associations between physical multimorbidity patterns and common mental health disorders in middle-aged adults: A prospective analysis using data from the UK Biobank. Lancet Reg Health-Europe. 2021;8:100149. [DOI] [PMC free article] [PubMed]
- 4.Socal MP, Trujillo AJ. Links Between Chronic Illness and Late-Life Cognition: Evidence From Four Latin American Countries. J Aging Health. 2018;30(2):262–304. [DOI] [PubMed] [Google Scholar]
- 5.Wu Y, Niu X, Lv BS, Li XH, Wang WJ, Feng WJ, Duan HP, Wu YL. The impacts of multimorbidity trajectories and patterns on functional limitations over time in middle-aged and older adults. Arch Gerontol Geriatr. 2025;137:105919. [DOI] [PubMed]
- 6.Caspersen CJ, Powell KE, Christenson GM. PHYSICAL-ACTIVITY, EXERCISE, AND PHYSICAL-FITNESS - DEFINITIONS AND DISTINCTIONS FOR HEALTH-RELATED RESEARCH. Public Health Rep. 1985;100(2):126–31. [PMC free article] [PubMed] [Google Scholar]
- 7.Schuch FB, Vancampfort D, Firth J, Rosenbaum S, Ward PB, Silva ES, Hallgren M, De Leon AP, Dunn AL, Deslandes AC, et al. Physical Activity and Incident Depression: A Meta-Analysis of Prospective Cohort Studies. Am J Psychiatry. 2018;175(7):631–48. [DOI] [PubMed] [Google Scholar]
- 8.Mammen G, Faulkner G. Physical Activity and the Prevention of Depression A Systematic Review of Prospective Studies. Am J Prev Med. 2013;45(5):649–57. [DOI] [PubMed] [Google Scholar]
- 9.Teychenne M, Ball K, Salmon J. Sedentary Behavior and Depression Among Adults: A Review. Int J Behav Med. 2011;17(4):246–54. [DOI] [PubMed] [Google Scholar]
- 10.Lenharo M. The surprisingly big health benefits of just a little exercise. Nature. 2026;649(8099):1092–4. [DOI] [PubMed] [Google Scholar]
- 11.Kritz-Silverstein D, Barrett-Connor E, Corbeau C. Cross-sectional and prospective study of exercise and depressed mood in the elderly: the Rancho Bernardo study. Am J Epidemiol. 2001;153(6):596–603. [DOI] [PubMed] [Google Scholar]
- 12.Johnson CC, Murray DM, Elder JP, Jobe JB, Dunn AL, Kubik M, Voorhees C, Schachter K. Depressive symptoms and physical activity in adolescent girls. Med Sci Sports Exerc. 2008;40(5):818–26. [DOI] [PubMed] [Google Scholar]
- 13.Chamberlain CEW, Lindsay S, Smith BJ, Henson SS, Sweet CC, Levens SM. Improvements in physical activity and depression symptoms: an observational study of users of a multi-modal digital mental health platform. Front Digit Health. 2025;7:1394647. [DOI] [PMC free article] [PubMed]
- 14.de Sá VR, Padovan-Neto FE. Protective Role of Regular Physical Activity on Mental Health During the COVID-19 Pandemic. Psychiatr Q. 2026;97(1):105–23. [DOI] [PubMed]
- 15.Lim Y. Longitudinal association between consecutive moderate-to-vigorous physical activity and the risk of depression among depressed and non-depressed participants: A nationally representative cohort study. J Affect Disord. 2025;381:475–83. [DOI] [PubMed] [Google Scholar]
- 16.Sheikh MA, Vancampfort D, Stubbs B. Leisure time physical activity and future psychological distress: A thirteen year longitudinal population-based study. J Psychiatr Res. 2018;101:50–6. [DOI] [PubMed] [Google Scholar]
- 17.Wang JJ. The longitudinal relationship between leisure activities and depressive symptoms among older Chinese adults: an autoregressive cross-lagged analysis approach. BMC Public Health. 2024;24(1):763. [DOI] [PMC free article] [PubMed]
- 18.Gudmundsson P, Lindwall M, Gustafson DR, Östling S, Hällström T, Waern M, Skoog I. Longitudinal associations between physical activity and depression scores in Swedish women followed 32 years. Acta psychiatrica Scandinavica. 2015;132(6):451–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Hulsegge G, Looman M, Smit HA, Daviglus ML, van der Schouw YT, Verschuren WMM. Lifestyle Changes in Young Adulthood and Middle Age and Risk of Cardiovascular Disease and All-Cause Mortality: The Doetinchem Cohort Study. J Am Heart Association. 2016;5(1):e002432. [DOI] [PMC free article] [PubMed]
- 20.Mulder M, Ranchor AV, Sanderman R, Bouma J, van den Heuvel WJA. The stability of lifestyle behaviour. Int J Epidemiol. 1998;27(2):199–207. [DOI] [PubMed] [Google Scholar]
- 21.Bonekamp NE, Visseren FLJ, Cramer MJ, Dorresteijn JAN, van der Meer MG, Ruigrok YM, van Sloten TT, Teraa M, Geleijnse JM, Koopal C, et al. Long-term lifestyle change and risk of mortality and Type 2 diabetes in patients with cardiovascular disease. Eur J Prev Cardiol. 2024;31(2):205–13. [DOI] [PubMed] [Google Scholar]
- 22.Xu CJ, Cao Z. Cardiometabolic diseases, total mortality, and benefits of adherence to a healthy lifestyle: a 13-year prospective UK Biobank study. J Translational Med. 2022;20(1):234. [DOI] [PMC free article] [PubMed]
- 23.Zeng Y, Feng QS, Hesketh T, Christensen K, Vaupel JW. Survival, disabilities in activities of daily living, and physical and cognitive functioning among the oldest-old in China: a cohort study. Lancet. 2017;389(10079):1619–29. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Wang J, Chen C, Zhou JH, Xu ZN, Xu LJ, Li XW, Zhong ZC, Lv YB, Shi XM. Integrated healthy lifestyle even in late-life mitigates cognitive decline risk across varied genetic susceptibility. Nat Commun. 2025;16(1):539. [DOI] [PMC free article] [PubMed]
- 25.Zhou YG, Kivimäki M, Yan LL, Carrillo-Larco RM, Zhang Y, Cheng YY, Wang H, Zhou MG, Xu XL. Associations between socioeconomic inequalities and progression to psychological and cognitive multimorbidities after onset of a physical condition: a multicohort study. Eclinicalmedicine. 2024;74:102739. [DOI] [PMC free article] [PubMed]
- 26.Ni YJ, Zhou YG, Kivimaeki M, Cai Y, Carrillo-Larco RM, Xu X, Dai XC, Xu XL. Socioeconomic inequalities in physical, psychological, and cognitive multimorbidity in middle-aged and older adults in 33 countries: a cross-sectional study. Lancet Healthy Longev. 2023;4(11):E618–28. [DOI] [PubMed] [Google Scholar]
- 27.Du M, Liu M, Liu J. Effects of physical and psychological multimorbidity on the risk of dementia: multinational prospective cohorts and a meta-analysis. BMC Med. 2024;22(1):423. [DOI] [PMC free article] [PubMed]
- 28.Hu CX, Jiang KX, Sun XY, He Y, Li RH, Chen YN, Zhang Y, Tao YC, Jin LN. Change in Healthy Lifestyle and Subsequent Risk of Cognitive Impairment Among Chinese Older Adults: A National Community-Based Cohort Study. Journals Gerontol Ser a-Biological Sci Med Sci. 2024;79(8):glae148. [DOI] [PubMed]
- 29.Ardebili HE, Khosravi S, Larijani B, Nedjat S, Nasrabadi AN. Psychometric evaluation of the persian version of the ‘aging male scales’ questionnaire. Int J Prev Med. 2014;5(9):1178. [PMC free article] [PubMed] [Google Scholar]
- 30.Kroenke K, Spitzer RL, Williams JBW. The Patient Health Questionnaire-2 - Validity of a two-item depression screener. Med Care. 2003;41(11):1284–92. [DOI] [PubMed] [Google Scholar]
- 31.Huang JH, Wang XH. Association of depressive symptoms with risk of incidence low back pain in middle-aged and older Chinese adults. J Affect Disord. 2024;354:627–33. [DOI] [PubMed] [Google Scholar]
- 32.Dong Y, Huang JH, Liu HQ. Independent and joint associations of dietary diversity and physical activity on mental health among older adults in China: a cross-sectional study. BMC Public Health. 2025;25(1):599. [DOI] [PMC free article] [PubMed]
- 33.Huang JH, Peng DR, Wang XH. Estimating the impact of metabolic syndrome on low back pain and the joint effects of metabolic syndrome and depressive symptoms on low back pain: insights from the China Health and Retirement Longitudinal Study. BMC Public Health. 2024;24(1):2359. [DOI] [PMC free article] [PubMed]
- 34.He XJ, Song XR, Liang JY, Ouyang YX, Gao YY, Sun XP, Feng Y, Ding QL, Zhen Q, Kou CG et al. The association between healthy lifestyle trajectory and depressive symptoms: Findings from the Chinese Longitudinal Healthy Longevity Survey (CLHLS). J Affect Disord. 2026;392:120170. [DOI] [PubMed]
- 35.Long QY, Li JY, Zhao N, Guo RW, Tang SL. Deep learning for depression prediction in older adults: A retrospective cohort study from CHARLS (2011–2020) with independent cohort validation in CLHLS (2008–2018). J Affect Disord. 2026;400:121206. [DOI] [PubMed]
- 36.Yan LL, Li C, Zou S, Li Y, Gong E, He Z, Shao S, Jin X, Hua Y, Gallis JA, et al. Healthy eating and all-cause mortality among Chinese aged 80 years or older. Int J Behav Nutr Phys Act. 2022;19(1):60. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Nagin DS. Group-based Modeling of Development. Harvard University Press; 2009. https://www.degruyterbrill.com/document/doi/10.4159/9780674041318/html#bibliographicData.
- 38.Nagin DS, Odgers CL. Group-Based Trajectory Modeling in Clinical Research. Annu Rev Clin Psychol.2010;6:109–138. [DOI] [PubMed]
- 39.Huang JH, Xu TW, Dai Y, Li YP, Tu RP. Age-related differences in the number of chronic diseases in association with trajectories of depressive symptoms: a population-based cohort study. BMC Public Health. 2024;24(1):2496. [DOI] [PMC free article] [PubMed]
- 40.Huang JH, Chen L, Chen HY, Liu QD, Tu CD, Dai Y, Li YP, Tu RP. Abnormal blood biomarkers and trajectories of depressive symptoms among Chinese middle-aged and older adults. J Psychosom Res. 2025;191:112052. [DOI] [PubMed]
- 41.Schafer JL. Multiple imputation: a primer. Stat Methods Med Res. 1999;8(1):3–15. [DOI] [PubMed] [Google Scholar]
- 42.Ambrosetti M, Abreu A, Corrà U, Davos CH, Hansen D, Frederix I, Iliou MC, Pedretti RFE, Schmid JP, Vigorito C, et al. Secondary prevention through comprehensive cardiovascular rehabilitation: From knowledge to implementation. 2020 update. A position paper from the Secondary Prevention and Rehabilitation Section of the European Association of Preventive Cardiology. Eur J Prev Cardiol. 2021;28(5):460–95. [DOI] [PubMed] [Google Scholar]
- 43.Valenzuela PL, Ruilope LM, Santos-Lozano A, Wilhelm M, Kränkel N, Fiuza-Luces C, Lucia A. Exercise benefits in cardiovascular diseases: from mechanisms to clinical implementation. Eur Heart J. 2023;44(21):1874–89. [DOI] [PubMed] [Google Scholar]
- 44.Xup Y, Lup J, Lip M, Wangp T, Wang K, Cao QY, Ding Y, Xiang Y, Wang SY, Yang QQ, et al. Diabetes in China part 2: prevention, challenges, and progress. Lancet Public Health. 2024;9(12):e1098–104. [DOI] [PubMed] [Google Scholar]
- 45.Xu XL, Mishra GD, Jones M. Evidence on multimorbidity from definition to intervention: An overview of systematic reviews. Ageing Res Rev. 2017;37:53–68. [DOI] [PubMed] [Google Scholar]
- 46.Jeong SW, Kim SH, Kang SH, Kim HJ, Yoon CH, Youn TJ, Chae IH. Mortality reduction with physical activity in patients with and without cardiovascular disease. Eur Heart J. 2019;40(43):3547–55. [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
Supplementary Material 1. Table S1. Scoring standard of physical activity. Table S2. Scales and scoring standard of depressive symptoms. Table S3. Definition of covariates. Table S4. Definition of smoking, drinking, and diet. Table S5. Goodness-of-fit statistics of group-based trajectory analysis. Table S6. Sensitivity analysis of multiple imputation of association between trajectories of physical activity and depressive symptoms among older adults with chronic disease.
Data Availability Statement
All datasets used in this analysis are available for download at (https://opendata.pku.edu.cn/dataverse/CHADS).