Latent multimorbidity patterns and their longitudinal associations with depressive symptom trajectories and ADL limitations among middle-aged and older adults in China: a longitudinal analysis of the CHARLS

Latent multimorbidity patterns and their longitudinal associations with depressive symptom trajectories and ADL limitations among middle-aged and older adults in China: a longitudinal analysis of the CHARLS | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Latent multimorbidity patterns and their longitudinal associations with depressive symptom trajectories and ADL limitations among middle-aged and older adults in China: a longitudinal analysis of the CHARLS Kaige Gao, Yanjie Zhao, Xuehong Zhang, Kanbao Su, Jia Jia, Yuezhen Xu This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9085989/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 11 You are reading this latest preprint version Abstract Background Multimorbidity is heterogeneous, but longitudinal evidence from China on how specific disease patterns relate to changes in depressive symptoms and functional limitations remains limited. This study aimed to identify baseline multimorbidity patterns and examine their longitudinal associations with trajectories of depressive symptoms and activities of daily living (ADL) limitations in a nationally representative cohort. Methods Data were drawn from the China Health and Retirement Longitudinal Study (CHARLS), 2011–2018. Latent class analysis (LCA) was used to identify baseline (2011) multimorbidity patterns based on 12 chronic conditions. Linear and generalized linear mixed-effects models were applied to assess the associations between baseline patterns and changes in depressive symptoms and the risk of ADL limitation from 2013 to 2018. Exploratory mediation analysis examined whether pain statistically mediated the association between multimorbidity patterns and subsequent depressive symptoms. Results Four multimorbidity patterns were identified: cardiometabolic (Class 1), respiratory-dominant (Class 2), musculoskeletal–digestive (Class 3), and relatively healthy (Class 4). Compared with Class 4, Class 3 showed a steeper increase in depressive symptom scores over time (β = 0.099, P = 0.020). Findings were directionally consistent in sensitivity analyses, including the random-slope model (β = 0.096) and the high-classification-certainty subsample (β = 0.102). In exploratory mediation analyses, pain indicators measured in 2013 were associated with higher subsequent depressive scores, but no clear statistically indirect effect was observed for Class 3. The risk of ADL limitation increased over time (P 0.05). Conclusions Distinct multimorbidity patterns were differentially associated with depressive symptom trajectories among middle-aged and older adults in China. The musculoskeletal–digestive pattern was associated with a faster increase in depressive symptoms over time. Pain may reflect an important symptom-burden correlate of this pattern, but no clear mediation evidence was observed under the current model specification. These findings suggest that mental health screening and symptom-oriented assessment may be particularly relevant for high-symptom-burden multimorbidity patterns in primary care. Middle-aged and older adults Multimorbidity Latent class analysis Depressive symptoms Activities of daily living China Figures Figure 1 Figure 2 1. Introduction Population aging is accelerating worldwide, alongside a growing burden of chronic disease among middle-aged and older adults. Multimorbidity has become an increasing challenge for healthcare systems, particularly for care coordination, long-term care planning, and resource allocation[ 1 , 2 ]. In addition to disease accumulation, multimorbidity is often accompanied by greater symptom burden, polypharmacy, and a higher risk of drug–drug interactions, and has been associated with increased healthcare use, functional decline, and poorer health-related quality of life[ 2 ]. In healthy aging and chronic disease management, identifying high-risk groups and understanding how psychological and functional outcomes change over time may help improve risk stratification, intervention planning, and resource allocation[ 3 ]. Multimorbidity is not a homogeneous construct. In epidemiological research, it is commonly defined using a threshold approach (e.g., ≥ 2 chronic conditions) or simple disease counts. Although these approaches are useful for quantifying burden, they may overlook differences in how chronic conditions cluster and in the clinical implications of specific combinations[ 4 – 6 ]. Evidence suggests that chronic conditions tend to cluster into relatively stable patterns, such as cardiometabolic or respiratory–musculoskeletal groupings[ 7 ]. These patterns may reflect different underlying mechanisms and risk profiles and may therefore be associated with different trajectories of mental and functional outcomes. From a public health perspective, examining specific multimorbidity patterns may be more informative than relying on disease counts alone when assessing patient-centered outcomes. Latent class analysis (LCA) is a data-driven method for identifying such patterns. By modelling multiple binary disease indicators simultaneously, LCA classifies individuals into latent subgroups with similar co-occurrence structures and provides probabilistic class assignments, allowing the identification of interpretable multimorbidity patterns[ 8 ]. Although the use of LCA in multimorbidity research has increased in recent years, most studies remain cross-sectional. In longitudinal research, analyses often focus on a single outcome or a single time point, and less is known about whether baseline multimorbidity patterns are associated with different changes in psychological and functional outcomes over time[ 3 ]. Evidence linking baseline multimorbidity patterns to subsequent outcome trajectories therefore remains limited. This question is particularly relevant in China, where population aging is occurring rapidly. National surveys indicate that multimorbidity among adults aged 60 years and older is common, with reported prevalence estimates of approximately 46–57%, and the burden continues to increase[ 9 , 10 ]. These patterns reflect not only the epidemiological transition but also persistent urban–rural disparities and inequalities in healthcare access. Despite the scale of the burden, longitudinal evidence on how distinct multimorbidity patterns relate to mental and functional outcomes in the Chinese population remains limited [ 11 ]. Depressive symptoms and limitations in activities of daily living (ADL) are two indicators of aging-related vulnerability. Depressive symptoms are commonly measured using standardized instruments and are associated with healthcare use and mortality risk. ADL limitation reflects impairment in basic self-care and is a predictor of disability progression and long-term care needs[ 12 ]. Previous longitudinal studies suggest that psychological distress and functional decline may influence one another over time, but it remains unclear whether different multimorbidity structures are associated with different trajectories of these outcomes in the Chinese context[ 12 , 13 ]. Using nationally representative data from the China Health and Retirement Longitudinal Study (CHARLS), this study aimed to identify latent multimorbidity patterns at baseline (2011) and examine their longitudinal associations with subsequent changes in depressive symptoms and ADL limitation from 2013 to 2018. Specifically, we aimed to: (1) use LCA to identify latent classes based on baseline chronic disease indicators; (2) assess differences in depressive symptom levels and the risk of ADL limitation across classes during follow-up; and (3) examine whether rates of change in these outcomes differed across classes through class-by-time interaction terms. 2. Methods 2.1 Study design and sample This prospective observational study examined whether baseline multimorbidity patterns were associated with subsequent longitudinal outcomes. Data came from the China Health and Retirement Longitudinal Study (CHARLS), a nationally representative panel survey of Chinese adults aged 45 years and older conducted by the National School of Development at Peking University[ 14 ]. Across repeated survey waves, CHARLS collects information on demographic characteristics, socioeconomic status, health conditions, and functional status. For the present analysis, the 2011 wave served as baseline, and follow-up information was taken from the 2013, 2015, and 2018 waves. The dataset is publicly available upon approved application. At baseline, CHARLS included 17,708 respondents. Because latent class membership could not be estimated for participants with missing data on all 12 chronic disease indicators, these individuals were excluded. This left 17,091 participants with non-missing most-likely latent class assignments for the latent class analysis (LCA) (Table 3 ). We further excluded respondents with implausible baseline age values and those who had no valid outcome observations during follow-up from 2013 to 2018. After these exclusions, the final longitudinal cohort comprised 15,879 individuals. Among them, 14,725 contributed to the depressive symptom models and 13,673 to the ADL limitation models. The difference in analytic sample size mainly reflected outcome-specific missingness and attrition across waves. Table 3 Estimated and Most Likely Latent Class Proportions (N = 17,091) Latent Class Model-estimated Proportion Most-likely Class Proportion Clinical Interpretation Class 4 69.11% 77.10% Relatively healthy Class 1 13.82% 10.58% Cardiometabolic Class 3 14.12% 10.01% Musculoskeletal–digestive Class 2 2.95% 2.32% Respiratory-dominant Note: N denotes the number of participants with non-missing most-likely class assignments obtained from the baseline latent class analysis (LCA). Model parameters in the LCA were estimated using full information maximum likelihood (FIML) to account for partial missing data in the chronic disease indicators. 2.2 Variable measurement and definitions 2.2.1 Exposure: Chronic disease assessment and latent class identification Baseline multimorbidity was defined using 12 chronic conditions assessed in 2011 on the basis of self-reported physician diagnoses or medical history. The conditions were hypertension, dyslipidemia, diabetes, malignant tumour, chronic lung disease, liver disease, heart disease, stroke, kidney disease, digestive disease, arthritis, and asthma. In large epidemiological surveys, self-reported physician diagnoses are commonly used. Studies based on CHARLS have shown moderate agreement and high specificity when these reports are compared with objective biomedical measurements, supporting their use in population-based research [ 15 , 16 ]. To identify multimorbidity patterns, we conducted LCA in Mplus version 8.3 using the 12 binary chronic disease indicators. Models with two to six classes were estimated. Selection of the final model considered several criteria: the Akaike Information Criterion (AIC), Bayesian Information Criterion (BIC), sample-size adjusted BIC (aBIC), entropy, the Lo–Mendell–Rubin adjusted likelihood ratio test (LMR), the bootstrap likelihood ratio test (BLRT), class size, and clinical interpretability. Within the LCA framework, missing data on individual disease indicators were handled using full information maximum likelihood (FIML), so all available responses could contribute to estimation. Participants with missing values on all 12 indicators were excluded because class membership could not be estimated. Once the final model had been selected, participants were assigned to latent classes using the most-likely class approach based on maximum posterior probability (MaxPP). Class membership was then entered into subsequent regression analyses as a categorical exposure variable, with the relatively healthy class as the reference group. Posterior probabilities were retained to evaluate classification quality. To assess the possible influence of classification uncertainty, we repeated key analyses in subsamples restricted to participants with higher assignment certainty according to prespecified MaxPP thresholds. 2.2.2 Outcomes Depressive symptoms were measured with the 10-item Center for Epidemiologic Studies Depression Scale (CES-D-10). This instrument has been validated in Chinese populations and has shown acceptable reliability and construct validity[ 17 ]. Scores range from 0 to 30, with higher values indicating more severe depressive symptoms. Before summation, the two positively worded items were reverse-coded. For the primary longitudinal analyses, CES-D-10 total score was modelled as a continuous outcome. A conventional cut-off of ≥ 10 was used for descriptive purposes to indicate elevated depressive symptoms, but it was not adopted in the primary modelling strategy unless otherwise specified. ADL limitation was defined as reporting difficulty or requiring assistance in at least one basic activity of daily living, namely dressing, bathing, eating, getting in or out of bed, toileting, and continence. At each follow-up wave, this information was used to construct a binary variable (any limitation vs. none). This variable was then used to examine change in the risk of ADL limitation over time. 2.2.3 Covariates All covariates were measured at baseline in 2011 and treated as time-invariant in the longitudinal analyses. They included age, sex, educational attainment, marital status, urban–rural residence, body mass index (BMI), and smoking status. Because baseline levels of the outcomes may influence subsequent change, the models also adjusted for the corresponding baseline outcome measure. Specifically, the depressive symptom models included the 2011 CES-D score, whereas the ADL models included baseline ADL limitation status. This adjustment was intended to reduce confounding by initial health status. 2.3 Statistical analysis To analyse the longitudinal associations between baseline multimorbidity patterns and subsequent outcomes, we constructed a person-wave dataset in long format and fitted mixed-effects models. Depressive symptoms (CES-D-10 total score) were analysed using linear mixed-effects models (LMMs), whereas ADL limitation was analysed using logistic generalized linear mixed-effects models (GLMMs). In all models, a random intercept was included to account for between-individual differences in baseline outcome levels. The main parameter of interest was the interaction between latent class membership and time, which tested whether rates of change in the outcomes differed across multimorbidity patterns. Time was parameterized as years since the first post-baseline follow-up wave (2013) and coded 0, 2, and 5 for the 2013, 2015, and 2018 waves, respectively. Exposure, covariates, and baseline outcome levels were all defined from the 2011 wave, which was therefore not included in the trajectory models. Model adjustment covered baseline demographic characteristics, socioeconomic factors, health behaviours, and the corresponding baseline outcome measure. Missing baseline covariate data were addressed using multiple imputation by chained equations (MICE; m = 20), after which estimates were pooled according to Rubin’s rules [ 18 , 19 ]. The LCA was performed before imputation based on the baseline chronic disease indicators, and missingness in those indicators was handled in Mplus using FIML. For that reason, the LCA indicators themselves were not imputed. Missing follow-up outcome data were handled under the maximum likelihood framework of the mixed-effects models, assuming missing at random (MAR). Sensitivity analyses considered alternative parameterizations of time, random-slope models, and analyses restricted to outcome-free subsamples at baseline. In addition, exploratory mediation analysis was conducted within a counterfactual framework to estimate the statistical indirect effect of pain on subsequent depressive symptoms. 2.4 Robustness and classification certainty analyses Several sensitivity analyses were performed to examine whether the primary findings were robust to alternative model specifications and sample restrictions. We first considered alternative ways of modelling time. To allow for potential non-linear trends, the continuous time variable was replaced with categorical wave indicators, using 2013 as the reference. In the depressive symptom models, we also fitted random-slope models for time to account for between-individual variation in rates of change. We next repeated the primary analyses in restricted subsamples to reduce the possibility that baseline outcome status influenced subsequent associations. For depressive symptoms, analyses were limited to participants without elevated depressive symptoms at baseline (CES-D < 10). For ADL limitation, analyses were restricted to those without baseline ADL limitation. These models were used to examine whether the observed associations were similar among participants free of the corresponding outcome at study entry. Classification certainty was examined in a further set of analyses. The primary models were based on most-likely class assignment. To assess whether classification uncertainty materially affected the estimates, we re-estimated the models after restricting the sample to participants with higher assignment certainty, defined by maximum posterior probability (MaxPP) thresholds of ≥ 0.70 and ≥ 0.80. 2.5 Mechanistic exploration: mediation analysis To explore a possible pathway underlying the association between the musculoskeletal–digestive pattern (Class 3) and subsequent depressive symptoms, we conducted an exploratory mediation analysis within a counterfactual framework following Imai et al[ 20 , 21 ]. Latent class membership in 2011 was treated as the exposure, pain indicators measured in 2013 as the mediator, and depressive symptom scores measured in 2015 and 2018 as the outcome. Pain was examined in three forms: any pain, pain severity, and number of pain sites. The primary comparison was between Class 3 and the relatively healthy class (Class 4). Mediator models were specified according to the type of mediator, using logistic regression for binary pain indicators and linear regression for continuous or count-based pain indicators. Outcome models were estimated using linear regression adjusted for baseline depressive symptoms, baseline demographic and health-behaviour covariates, and survey wave. The average causal mediation effect (ACME) and average direct effect (ADE) were estimated. For the multiply imputed datasets (MICE, m = 20), mediation analyses were conducted separately within each imputed dataset and pooled according to Rubin’s rules. Given the observational design and the assumptions required for mediation analysis, these results were interpreted as exploratory evidence relevant to potential mechanisms rather than as confirmation of a causal pathway. 3. Results 3.1 Identification of latent classes Using the 12 binary chronic disease indicators measured at baseline in 2011, we fitted latent class models with two to six classes and compared their fit (Table 1 ). Model selection considered statistical fit indices, likelihood ratio tests, entropy, class size, and clinical interpretability. On balance, the four-class model was selected for the primary analyses. Table 1 Model Fit Indices for Latent Class Analysis (K = 2–6) Classes (K) LL AIC BIC aBIC Entropy LMR P-value BLRT P-value 2 Classes -60820.794 121,691.59 121,885.25 121,805.80 0.543 < 0.001 < 0.001 3 Classes -60113.081 120,302.16 120,596.52 120,475.76 0.636 < 0.001 < 0.001 4 Classes -59816.962 119,735.93 120,130.99 119,968.91 0.702 0.001 < 0.001 5 Classes -59660.225 119,448.45 119,944.21 119,740.83 0.63 < 0.001 < 0.001 6 Classes -59619.025 119,392.05 119,988.52 119,743.82 0.647 0.046 < 0.001 Note: LL = log-likelihood; AIC = Akaike Information Criterion; BIC = Bayesian Information Criterion; aBIC = sample-size adjusted BIC. The four-class solution was selected as the final latent class model. The four-class model had an entropy of 0.702, higher than that of the three-class model (0.636) and the five-class model (0.630) (Table 1 ). Although the five-class model showed a slightly lower BIC than the four-class model, the BIC increased again in the six-class model. In addition, models with more classes tended to yield smaller class sizes and lower classification quality. Taken together, these results supported the four-class solution as the most interpretable and parsimonious model. The LMR and BLRT also indicated improved fit up to the four-class model (Table 1 ). Conditional probabilities from the four-class model suggested four distinct multimorbidity patterns (Table 2 and Fig. 1 ). Class 1 was labelled cardiometabolic, characterized by higher probabilities of hypertension, dyslipidemia, and diabetes. Class 2 was labelled respiratory-dominant, with chronic lung disease and asthma as the main conditions. Class 3 was labelled musculoskeletal–digestive, marked by arthritis and stomach or digestive disease. Class 4 was labelled relatively healthy, with low probabilities across all conditions. In the five- and six-class models, additional classes mainly reflected further subdivision of these patterns or smaller subgroups with limited added clinical interpretability. Table 2 Item-Response Probabilities for the Four-Class Model Chronic Conditions Cardiometabolic Respiratory-dominant Musculoskeletal–digestive Relatively healthy hypertension 0.784 0.277 0.288 0.133 dyslipidemia 0.411 0.103 0.111 0.027 diabetes mellitus 0.261 0.067 0.055 0.018 heart disease 0.363 0.294 0.284 0.033 stroke 0.109 0.034 0.026 0.006 chronic lung disease 0.071 1 0.233 0.045 asthma 0.023 0.702 0.05 0.009 arthritis 0.313 0.443 0.798 0.237 stomach or digestive disease 0.148 0.293 0.62 0.154 kidney disease 0.078 0.101 0.221 0.028 liver disease 0.045 0.056 0.133 0.018 cancer 0.017 0 0.016 0.008 Note: Values represent the item–response (conditional) probabilities estimated from the four-class latent class model. Probabilities indicate the likelihood of reporting the presence of each chronic condition (coded as 1) given membership in a specific latent class. Higher values indicate greater concentration of the condition within that class. These probabilities are model estimates and do not represent crude prevalence in the overall sample. As shown in Table 3 , the model-estimated class proportions and the most-likely class assignments were broadly similar. Because some degree of classification uncertainty remained, we conducted additional analyses using maximum posterior probability (MaxPP) thresholds of ≥ 0.70 and ≥ 0.80. Key class-by-time interaction terms were then re-estimated in subsamples with higher assignment certainty to examine whether the direction and magnitude of the estimates changed materially. 3.2 Baseline characteristics across multimorbidity patterns Among the 15,879 participants included in the longitudinal cohort, several baseline characteristics differed across the four latent classes (Table 4 ). Age, sex, and marital status did not differ significantly across classes (all P > 0.05). In contrast, educational attainment, urban–rural residence, and BMI showed significant between-class differences (all P < 0.01). Participants in the cardiometabolic class (Class 1) had the highest mean BMI. Table 4 Baseline characteristics across latent multimorbidity classes among participants included in the longitudinal cohort (N = 15,879) Characteristic Relatively healthy (n = 12 ,249) Cardiometabolic (n = 1 ,669) Respiratory-dominant (n = 362) Musculoskeletal–digestive (n = 1 ,599) P value Age (mean (SD)) 58.82 (9.66) 58.49 (9.20) 58.73 (9.16) 58.94 (9.73) 0.554 Female, n (%) 6315 (51.6) 865 (51.8) 172 (47.5) 816 (51.1) 0.467 Education level, % 0.006 No formal education 3346 (27.4) 418 (25.1) 104 (28.7) 450 (28.2) Primary school 4868 (39.8) 625 (37.5) 131 (36.2) 664 (41.6) Middle school 2496 (20.4) 391 (23.5) 83 (22.9) 311 (19.5) High school and above 1524 (12.5) 232 (13.9) 44 (12.2) 173 (10.8) Residence (Urban), % 5641 (46.1) 809 (48.5) 172 (47.5) 734 (46.0) < 0.001 Marital status (Married), % 10717 (87.5) 1484 (88.9) 317 (87.6) 1388 (86.8) 0.301 Current smoking, % 3494 (29.6) 464 (28.7) 108 (30.9) 445 (28.7) 0.698 BMI (mean (SD)) 23.48 (3.93) 24.21 (4.37) 23.35 (3.61) 23.45 (3.91) < 0.001 Baseline CES-D-10 score, mean (SD) 8.45 (6.34) 8.17 (6.53) 8.46 (6.55) 8.75 (6.22) 0.106 No ADL limitation, % 6408 (76.2) 894 (77.7) 202 (78.0) 886 (75.0) 0.4 Note: Continuous variables are presented as mean (standard deviation, SD), and categorical variables as number (percentage, %). Baseline depressive symptoms were assessed using the CES-D-10 scale (range: 0–30), with higher scores indicating more severe symptoms. Limitation in activities of daily living (ADL) was defined as dependence on or need for assistance with at least one basic ADL task. P-values for continuous variables were derived from one-way analysis of variance (ANOVA), and P-values for categorical variables from chi-square tests. The total sample entering the longitudinal analyses was N = 15,879. Sample sizes varied slightly across outcome models due to missing data; details are provided in the Methods section. The crude distribution of individual chronic conditions across classes was generally consistent with the conditional probabilities estimated from the LCA model (Table 4 ). Baseline CES-D-10 scores and the prevalence of ADL limitation were similar across classes, with no statistically significant differences observed (both P > 0.05). These baseline characteristics provide context for the subsequent longitudinal analyses, in which all models were adjusted for baseline covariates and the corresponding baseline outcome measure. 3.3 Longitudinal associations between multimorbidity patterns and depressive symptom trajectories In linear mixed-effects models pooled across 20 imputed datasets, CES-D-10 scores increased over follow-up from 2013 to 2018 (β = 0.200 per year, 95% CI: 0.171–0.229; P < 0.001) (Table 5 ). Adjusted predicted trajectories are shown in Fig. 2 . Compared with the relatively healthy class (Class 4), the musculoskeletal–digestive class (Class 3) was the only class with a significant class-by-time interaction (β = 0.099, 95% CI: 0.016–0.183; P = 0.020), corresponding to a steeper increase in depressive symptom scores over time. Baseline depressive symptom scores did not differ significantly between Class 3 and Class 4 (P = 0.538). For the cardiometabolic class (Class 1), the class-by-time interaction was positive but did not reach statistical significance (P = 0.083). Table 5 Longitudinal Associations Between Multimorbidity Patterns and Depressive Symptom Trajectories (CES-D Scores) Among Middle-Aged and Older Adults, 2013–2018 Variable β (95% CI) P value Baseline level, vs. Class 4 (Relatively healthy) Cardiometabolic (Class 1) -0.142 (-0.459, 0.176) 0.382 Respiratory-dominant (Class 2) 0.128 (-0.499, 0.754) 0.689 Musculoskeletal–digestive (Class 3) 0.101 (-0.221, 0.423) 0.538 Time effect (per year) 0.200 (0.171, 0.229) < 0.001 Interaction terms: differences in rate of change Cardiometabolic × Time 0.074 (-0.010, 0.157) 0.083 Respiratory-dominant × Time 0.037 (-0.127, 0.201) 0.658 Musculoskeletal–digestive × Time 0.099 (0.016, 0.183) 0.020 Note: CES-D = Center for Epidemiologic Studies Depression Scale; CI = confidence interval. All estimates were obtained from linear mixed-effects models (LMMs) and pooled across 20 multiply imputed datasets using Rubin’s rules. Interaction terms represent the estimated differences in the rate of change in depressive symptoms over time for each multimorbidity pattern relative to the reference group. Statistical significance was defined as P < 0.05. 3.4 Longitudinal associations between multimorbidity patterns and risk of ADL limitation In logistic generalized linear mixed-effects models pooled across 20 imputed datasets, the odds of ADL limitation increased over time (OR = 1.083 per year, 95% CI: 1.064–1.103; P < 0.001) (Table 6 ). After adjustment for baseline ADL limitation status and covariates, the musculoskeletal–digestive class (Class 3: OR = 1.141, 95% CI: 0.959–1.358; P = 0.138) and the cardiometabolic class (Class 1: OR = 1.136, 95% CI: 0.950–1.359; P = 0.162) showed higher point estimates than the relatively healthy class, although these differences were not statistically significant. Table 6 Longitudinal Associations Between Multimorbidity Patterns and Risk of ADL Limitation Among Middle-Aged and Older Adults, 2013–2018 Variable OR (95% CI) P value Baseline level, vs. Class 4 (Relatively healthy) Cardiometabolic (Class 1) 1.136 (0.950, 1.359) 0.162 Respiratory-dominant (Class 2) 0.949 (0.659, 1.367) 0.780 Musculoskeletal–digestive (Class 3) 1.141 (0.959, 1.357) 0.138 Time effect (per year) 1.083 (1.064, 1.103) < 0.001 Interaction terms: differences in rate of change Cardiometabolic × Time 1.029 (0.978, 1.082) 0.268 Respiratory-dominant × Time 1.029 (0.930, 1.140) 0.576 Musculoskeletal–digestive × Time 0.984 (0.936, 1.033) 0.515 Note: ADL = activities of daily living; OR = odds ratio; CI = confidence interval. Estimates were derived from generalized linear mixed-effects models (GLMMs) and pooled across 20 multiply imputed datasets using Rubin’s rules. Class 4 (relatively healthy) served as the reference group. Interaction terms represent differences in the rate of change in ADL limitation risk over time between each multimorbidity pattern and the reference group. Statistical significance was defined as P 0.05), suggesting that the rate of change in ADL limitation risk did not differ across multimorbidity classes under the specified model. 3.5 Robustness and sensitivity analyses Sensitivity analyses were conducted to examine the robustness of the primary findings under alternative model specifications and sample restrictions (Table 7 ). Among participants without elevated depressive symptoms at baseline, the class-by-time interaction for the musculoskeletal–digestive class (Class 3) was similar in magnitude to that in the primary analysis (β = 0.095 vs. 0.099), although the estimate was less precise in the smaller sample (P = 0.059). In random-slope models that allowed individual-specific rates of change, the Class 3 interaction remained similar to the primary estimate (β = 0.096; P = 0.028). When time was parameterized using wave indicators, depressive symptoms increased across follow-up waves, and the difference between Class 3 and the relatively healthy class was larger at the 2018 wave (β for Class 3 vs. Class 4 at 2018 relative to 2013 = 0.496; P = 0.020), consistent with the positive class-by-time interaction observed in the primary model. Table 7 Summary of Robustness Analyses for the Association Between the Musculoskeletal–Digestive Multimorbidity Pattern and Health Outcomes Robustness analysis model Depressive symptoms (DEP) β (95% CI) P value ADL limitation OR (95% CI) P value Primary analysis model 0.099 (0.016, 0.183) 0.020 0.984 (0.936, 1.033) 0.515 Nonlinear time specification 0.496 (0.077, 0.915) 0.020 0.933 (0.725, 1.200) 0.589 Random-slope model 0.096 (0.010, 0.182) 0.028 0.984 (0.937, 1.034) 0.519 Baseline outcome exclusion 0.095 (-0.004, 0.193) 0.059 0.983 (0.925, 1.043) 0.564 High classification certainty sample 0.102 (-0.030, 0.234) 0.131 0.994 (0.921, 1.072) 0.874 Note: DEP = depressive symptoms; ADL = activities of daily living. This table summarizes the interaction coefficients between the musculoskeletal–digestive multimorbidity pattern (Class 3) and time (or survey wave) across multiple robustness analyses. All models were adjusted for the same covariates as in the primary analysis. For ADL outcomes, the reported odds ratios correspond to the Class 3 × Time interaction term, representing differences in the rate of change in ADL limitation risk over the follow-up period rather than baseline group differences. In analyses restricted to participants with higher classification certainty, the Class 3 interaction estimates in the depressive symptom models were broadly similar to those in the primary analysis (e.g., β = 0.102), although confidence intervals widened in some specifications because of smaller sample size. In the ADL limitation models, point estimates for some classes increased as the maximum posterior probability (MaxPP) threshold was raised (e.g., OR ≈ 1.26), but the confidence intervals continued to include the null. Overall, the direction of the estimates was generally consistent across sensitivity analyses. Although statistical significance varied across some specifications, the results did not materially differ from the primary findings. 3.6 Exploratory analysis of pain distribution and mediation Because the musculoskeletal–digestive class (Class 3) showed a steeper increase in depressive symptoms over follow-up, we further examined pain characteristics measured in 2013 and their potential relevance to later depressive symptoms. As shown in Table 8 , overall pain prevalence, pain severity, and number of pain sites differed across classes. Class 3 had the highest overall pain prevalence (40.2%). It also showed a heavier overall pain burden in terms of pain severity and number of pain sites (both P < 0.05). By contrast, no significant between-class differences were observed for single-site pain indicators, including stomach, waist, and knee pain. Although the overall distributions of pain severity and number of pain sites differed across classes, their medians and interquartile ranges were the same, suggesting that the differences were more apparent in the distribution as a whole than in central tendency. Table 8 Pain characteristics in 2013 across latent multimorbidity classes (N = 12,585) Pain indicators Class 4 (Relatively healthy) (n = 9,698) Class 1 (Cardiometabolic) (n = 1,303) Class 2 (Respiratory) (n = 302) Class 3 (Musculoskeletal–digestive) (n = 1,282) P value Overall pain prevalence (%) 36.7 34.5 34.8 40.2 0.017 Site-specific pain prevalence (%) Stomach 4.1 3.9 3.0 5.3 0.127 Lower back 15.9 14.7 14.2 18.0 0.094 Knees 9.6 8.5 8.6 10.5 0.369 Number of pain sites, median [IQR] 0 [0,1] 0 [0,1] 0 [0,1] 0 [0,1] 0.017 Pain severity (Median [IQR]) 0 [0,1] 0 [0,1] 0 [0,1] 0 [0,1] 0.001 Note: Categorical variables are presented as percentages, and ordinal or count variables as median [IQR]. Between-class differences were assessed using chi-square tests for categorical variables and Kruskal–Wallis tests for pain severity and number of pain sites. The total sample (N = 12,585) included participants in the longitudinal cohort who had at least one depressive outcome record in 2015 or 2018 and had available pain measures in 2013. Although the median and IQR were identical across classes, the Kruskal–Wallis test indicated differences in the overall distributions. We then examined whether pain indicators measured in 2013 statistically mediated the association between baseline multimorbidity class and depressive symptom scores in 2015 and 2018. Across all three pain measures—any pain, pain severity, and number of pain sites—pain in 2013 was associated with higher subsequent depressive scores. However, the association between Class 3 and these pain indicators was modest, and no indirect effect reached statistical significance. For example, when number of pain sites was used as the mediator, the average causal mediation effect (ACME) was 0.032 (95% CI: −0.012 to 0.081; P = 0.190). Overall, these findings suggest that pain may reflect a higher symptom burden in the musculoskeletal–digestive class and may be related to subsequent depressive symptoms. Under the current model specification, however, no clear statistical mediation effect was observed. 4. Discussion Multimorbidity is increasingly recognized as a heterogeneous construct rather than a simple accumulation of chronic conditions[ 7 , 22 ]. Much of the existing literature has relied on disease counts or cross-sectional clustering, with less attention to how baseline multimorbidity patterns relate to subsequent psychological and functional outcomes over time[ 23 ]. In China, associations between multimorbidity patterns and depression or functional impairment have been reported, but longitudinal evidence remains limited[ 24 ]. Using a nationally representative cohort, we identified four clinically interpretable multimorbidity patterns and found that depressive symptom trajectories differed across classes, whereas the rate of change in ADL limitation did not. These findings support a structure-oriented perspective on multimorbidity and suggest that psychological and functional outcomes may not evolve in the same way across multimorbidity patterns[ 25 , 26 ]. The main finding of this study was that the musculoskeletal–digestive pattern (Class 3) was associated with a steeper increase in depressive symptoms over time, and this pattern was generally consistent across sensitivity analyses. Baseline depressive symptom levels did not differ significantly between Class 3 and the relatively healthy class, but their trajectories diverged during follow-up. This suggests that multimorbidity structure may be relevant not only to baseline burden, but also to subsequent change in mental health[ 27 ]. The finding is also consistent with previous studies showing that specific multimorbidity patterns are associated with later-life depression and poorer mental health outcomes[ 28 ]. One possible explanation is symptom burden. Persistent pain, gastrointestinal discomfort, and related symptoms may affect mood through daily disruption, reduced activity, and lower social participation[ 29 – 33 ]. By contrast, cardiometabolic conditions may be less symptomatically salient at earlier stages and may be managed through more standardized care pathways, which could partly attenuate their psychological impact[ 5 , 28 ]. Previous studies have linked musculoskeletal disease and chronic pain to depression[ 34 ], and musculoskeletal-dominant multimorbidity patterns have also been associated with poorer quality of life and mental health[ 28 , 32 ]. Our findings extend this literature by suggesting that the difference may lie in the trajectory of depressive symptoms over time rather than in baseline depressive levels alone[ 27 , 35 ]. More broadly, they are consistent with the view that different combinations of chronic conditions may reflect different levels of physiological burden, symptom experience, and care needs, which may in turn shape subsequent psychological outcomes[ 6 , 24 , 36 ]. The exploratory mediation analysis showed that pain was associated with higher subsequent depressive symptom scores, but no clear statistical indirect effect was observed for the musculoskeletal–digestive pattern under the current model specification. Pain therefore appears more consistent with a symptom-burden correlate than with a confirmed mediator in this analysis. Even so, pain may still be relevant to the association between this pattern and later depressive symptoms. Within a biopsychosocial framework, pain may influence mood through disrupted sleep and a possible pain–insomnia–depression cycle[ 33 , 37 ], reduced activity and social participation accompanied by greater isolation or perceived loss[ 27 ], and increased bodily vigilance related to gastrointestinal symptoms[ 31 ]. These possibilities are broadly consistent with the higher overall pain burden observed in the musculoskeletal–digestive class. Other mechanisms, including systemic inflammation and gut–brain pathways, have also been proposed in the literature, but they were not measured in the present study and therefore remain speculative[ 23 ]. Further studies incorporating inflammatory markers, sleep measures, medication use, and repeated pain assessments may help clarify the behavioural and biological pathways involved. Unlike the findings for depressive symptoms, we did not observe significant between-class differences in the rate of change in ADL limitation, although some point estimates were higher in the musculoskeletal–digestive and cardiometabolic classes. Several explanations are possible. First, change in basic ADL may emerge more slowly or nonlinearly than change in depressive symptoms, and the follow-up period may not have been long enough to detect differences in slopes[ 23 , 38 ]. Earlier or subtler functional changes may also be better captured by more sensitive measures such as instrumental activities of daily living (IADL)[ 38 ]. Second, attrition due to loss to follow-up or mortality may have attenuated associations, particularly if participants with poorer health were less likely to remain under observation. Classification uncertainty may also have reduced the contrast between classes, although point estimates tended to increase in some higher-certainty analyses[ 38 – 40 ]. Third, in community-dwelling populations, compensatory strategies may allow individuals to maintain basic ADL despite declines in intrinsic capacity[ 41 ]. Taken together, these findings suggest that multimorbidity structure may be more informative for depressive symptom trajectories than for short- to mid-term change in basic ADL. Within the present follow-up period, the increase in ADL limitation appeared to reflect a more general age-related pattern across classes. We also observed opposite directional associations of BMI with depressive symptom change and ADL limitation risk. This pattern is broadly consistent with discussions of an “obesity paradox” in older adults, whereby higher BMI may reflect greater nutritional reserve and lower wasting risk, while also increasing the risk of functional limitation through mechanical and cardiopulmonary burden[ 38 , 41 , 42 ]. At the same time, BMI is an imperfect marker in older populations because it does not distinguish fat mass from lean mass and may be influenced by illness-related weight loss or reverse causation. This finding should therefore be interpreted cautiously. Future studies incorporating body composition, sarcopenia, and physical performance measures may provide a clearer picture. From a public health and primary care perspective, these findings suggest that multimorbidity management may benefit from moving beyond simple disease counts. Stratification based only on the number of conditions may overlook clinically relevant heterogeneity. In particular, patterns characterized by higher symptom burden, such as the musculoskeletal–digestive class, may warrant closer attention to mental health risk and symptom management[ 43 , 44 ]. Several practical implications follow. First, mental health screening could be more explicitly incorporated into routine chronic disease management, particularly for individuals with chronic musculoskeletal pain and gastrointestinal conditions[ 28 ]. Second, care models that integrate symptom control and psychological support may be useful, especially where pain management, primary care, and mental health services are delivered separately[ 24 ]. Third, the absence of clear between-class differences in ADL slope, alongside the earlier divergence in depressive symptoms, may indicate a window for earlier intervention before functional decline becomes more apparent. In resource-limited settings, simple rule-based identification strategies and longitudinal follow-up within primary care or electronic health record systems may help target support more efficiently[ 23 , 43 – 45 ]. Limitations Several limitations should be considered when interpreting these findings. First, classification quality in the latent class model was moderate (entropy = 0.702), indicating some uncertainty in individual class assignment. Although sensitivity analyses using MaxPP thresholds showed generally similar results, the latent classes are better interpreted as population-level multimorbidity structures than as fixed labels at the individual level. Second, ADL limitation was self-reported and may have been less sensitive to early or subtle functional change. In addition, the respiratory-dominant class was relatively small, which reduced the precision of estimates for this pattern. Third, attrition due to loss to follow-up or mortality may have introduced selection bias if it was related to health status. Although missing data were handled using multiple imputation and maximum likelihood under a missing at random (MAR) assumption, missing not at random (MNAR) mechanisms cannot be ruled out. Finally, the exploratory mechanistic analyses were based on an observational design and remained subject to unmeasured confounding and other assumptions required for mediation analysis. These results should therefore be interpreted as exploratory evidence on possible pathways rather than confirmation of a causal mechanism. Future studies with longer follow-up, repeated symptom assessments, and more detailed behavioural and biological measures may help clarify these pathways. Conclusion Using nationally representative longitudinal data, this study identified four clinically interpretable multimorbidity patterns and showed that different multimorbidity structures were associated with different longitudinal patterns of depressive symptoms and ADL limitation. The musculoskeletal–digestive pattern was the only class associated with a faster increase in depressive symptoms over time. By contrast, although the risk of ADL limitation increased across all classes, no clear between-class differences in the rate of change were observed under the current analytical framework. Exploratory analyses further suggested that pain may reflect a higher symptom burden in the musculoskeletal–digestive pattern and may be related to subsequent depressive symptoms, although no clear mediation effect was observed. Overall, these findings support the value of considering multimorbidity structure alongside disease counts in risk assessment. In public health and primary care settings, identifying patterns with higher symptom burden may help inform mental health screening and follow-up among middle-aged and older adults. Abbreviations ADL Activities of daily living BMI Body mass index BLRT Bootstrap likelihood ratio test CES-D-10 10-item Center for Epidemiologic Studies Depression Scale CHARLS China Health and Retirement Longitudinal Study CI Confidence interval FIML Full information maximum likelihood GLMM Generalized linear mixed-effects model IADL Instrumental activities of daily living LCA Latent class analysis LMM Linear mixed-effects model LMR Lo–Mendell–Rubin adjusted likelihood ratio test MAR Missing at random MaxPP Maximum posterior probability MICE Multiple imputation by chained equations MNAR Missing not at random OR Odds ratio PHQ Patient Health Questionnaire SD Standard deviation Declarations Ethics approval and consent to participate The CHARLS study was approved by the Biomedical Ethics Review Committee of Peking University (IRB00001052-11015). The study was conducted in accordance with relevant institutional and national regulations. All participants provided written informed consent prior to participation. The study was conducted in accordance with the Declaration of Helsinki. Consent for publication Not applicable. Clinical trial number Not applicable. Competing interests The authors declare that they have no competing interests. Funding This study was supported by the Xinjiang Uygur Autonomous Region Social Science Fund Project 21BRK116. This study has also been approved as a general project of the Xinjiang Nursing Society in 2025, with project number 2025XH38. Author Contribution Kaige Gao, Yanjie Zhao, Xuehong Zhang, Kanbao Su, Jia Jia and Yuezhen Xu wrote the main manuscript text and prepared figures and tables. All authors reviewed the manuscript. Acknowledgement Thank Charls researchers and investigators for selflessly providing publicly available data Data Availability The datasets analyzed in this study are available in the China Health and Retirement Longitudinal Study (CHARLS) repository. Access to the data requires online registration and approval through the official CHARLS website. References Gulliford MC, Green JM. Is multimorbidity a useful concept for public health? Lancet Public Health. 2024;9(4):e210–1. Al-Chalabi S, Sinha S, Kalra PA. Enhancing clinical service design for multimorbidity management: A comprehensive approach to joined-up care for diabetes, chronic kidney disease, and heart failure. Diabet Med. 2025;42(2):e15403. Nguyen H, Moreno-Agostino D, Chua KC, Vitoratou S, Prina AM. Trajectories of healthy ageing among older adults with multimorbidity: A growth mixture model using harmonised data from eight ATHLOS cohorts. PLoS ONE. 2021;16(4):e0248844. Cezard G, McHale CT, Sullivan F, Bowles JKF, Keenan K. Studying trajectories of multimorbidity: a systematic scoping review of longitudinal approaches and evidence. BMJ Open. 2021;11(11):e048485. Zhang Y, Zhou Y, Kivimäki M, Holt-Lunstad J, Carrillo-Larco RM, Xu X. Cardiometabolic multimorbidity, social activity, and joint trajectories of physical disability, depressive symptom, and cognitive function in mid-to-late life: a multicohort study. BMC Med. 2025;23(1):577. Dong X, Ma Y, Zhang H, Wang P. Individual-level transitions between chronic disease multimorbidity clusters and the risk of five-year mortality in longitudinal cohort of Chinese middle-aged and older adults. Aging Clin Exp Res. 2025;37(1):216. Bayes-Marin I, Sanchez-Niubo A, Egea-Cortés L, Nguyen H, Prina M, Fernández D, Haro JM, Olaya B. Multimorbidity patterns in low-middle and high income regions: a multiregion latent class analysis using ATHLOS harmonised cohorts. BMJ Open. 2020;10(7):e034441. Sinha P, Calfee CS, Delucchi KL. Practitioner's Guide to Latent Class Analysis: Methodological Considerations and Common Pitfalls. Crit Care Med. 2021;49(1):e63–79. Liu Q, Lu J. Prevalence of multimorbidity and its relationship with socioeconomic status among Chinese older adults over time. PLoS ONE. 2025;20(4):e0322042. Hu X, Li S, Wei Z, Wu D, Meng L, Li J, Xu J, Zhang L, Ma Q, Li H, et al. Prevalence and pattern of multimorbidity in China: a cross-sectional study of 224,142 adults over 60 years old. Front Public Health. 2024;12:1349418. Long C, Huang J, Liu D, Liu C, Wu M, Wu H, Deng J, Zhang Y, Shi L, Cui Y. Prevalence, combination patterns, and quality of life factors of multimorbidity among older adults in southern China based on the health ecological model. J Glob Health. 2025;15:04215. Zhou L, Wang W, Ma X. The bidirectional association between the disability in activities of daily living and depression: a longitudinal study in Chinese middle-aged and older adults. BMC Public Health. 2024;24(1):1884. Zhu X, Wang Y, Luo Y, Ding R, Shi Z, He P. Bidirectional, longitudinal associations between depressive symptoms and IADL/ADL disability in older adults in China: a national cohort study. BMC Geriatr. 2024;24(1):659. Zhao Y, Hu Y, Smith JP, Strauss J, Yang G. Cohort profile: the China Health and Retirement Longitudinal Study (CHARLS). Int J Epidemiol. 2014;43(1):61–8. Ning M, Zhang Q, Yang M. Comparison of self-reported and biomedical data on hypertension and diabetes: findings from the China Health and Retirement Longitudinal Study (CHARLS). BMJ Open. 2016;6(1):e009836. Xie D, Wang J. Comparison of self-reports and biomedical measurements on hypertension and diabetes among older adults in China. BMC Public Health. 2020;20(1):1664. Zhao Y, Strauss J, Yang G, Giles J, Hu PP, Hu Y, Lei X, Liu M, Park A, Smith JP, et al. China Health and Retirement Longitudinal Study: 2011–2012 National Baseline User’s Guide. Beijing, China: National School of Development, Peking University; 2013. van Buuren S. Flexible Imputation of Missing Data. 2nd ed. New York: Chapman and Hall/CRC; 2018. Rubin DB. Multiple Imputation for Nonresponse in Surveys. New York: Wiley; 1987. Imai K, Keele L, Tingley D. A general approach to causal mediation analysis. Psychol Methods. 2010;15(4):309–34. Imai K, Keele L, Yamamoto T. Identification, Inference and Sensitivity Analysis for Causal Mediation Effects. Stat Sci. 2010;25(1):51–71. Ogaz-González R, Zou Q, Du Y, Gutiérrez-Robledo LM, Escamilla-Santiago R, López-Cervantes M, Corpeleijn E. Multimorbidity patterns in older adults from Northern Netherlands: comparing factor analysis and latent class analysis solutions. BMC Geriatr. 2025;25(1):381. Liu Q, Lin S, Yin L, Liu W, Zhu X, Li Z, Shen Y, Feng B. Identification and trajectory of multimorbidity patterns among older people in China: a longitudinal study based on the China health and retirement longitudinal study 2011–2020 data. Front Public Health. 2025;13:1597224. Bao L, Zhang W, Zhang Y, Wang M, Chen Y, Shan X, Xing Y. The relationships between multimorbidity, depressive symptoms, health service utilization, and activities of daily living among the elderly in China. PLoS ONE. 2025;20(10):e0333923. Wu Y, Su B, Chen C, Zhao Y, Zhong P, Zheng X. Urban-rural disparities in the prevalence and trends of depressive symptoms among Chinese elderly and their associated factors. J Affect Disord. 2023;340:258–68. Zheng DD, Loewenstein DA, Christ SL, Feaster DJ, Lam BL, McCollister KE, Curiel-Cid RE, Lee DJ. Multimorbidity patterns and their relationship to mortality in the US older adult population. PLoS ONE. 2021;16(1):e0245053. Ronaldson A, Arias de la Torre J, Bendayan R, Yadegarfar ME, Rhead R, Douiri A, Armstrong D, Hatch S, Hotopf M, Dregan A. Physical multimorbidity, depressive symptoms, and social participation in adults over 50 years of age: findings from the English Longitudinal Study of Ageing. Aging Ment Health. 2023;27(1):43–53. Wang X, Zheng N, Yin M. Multimorbidity Patterns and Depression: Bridging Epidemiological Associations with Predictive Analytics for Risk Stratification. Healthcare (Basel) 2025, 13(12). Lehti TE, Öhman H, Knuutila M, Kautiainen H, Karppinen H, Tilvis R, Strandberg T, Pitkälä KH. Symptom burden in community-dwelling older people: temporal trends in the Helsinki Aging Study. Aging Clin Exp Res. 2021;33(11):3065–71. McKay MA, Bernhardt P, O'Connor M, Leveille S. Characterizing symptom burden among community-dwelling older adults. Geriatr Nurs. 2026;69:103899. Huang MH, Wang YP, Wu PS, Chan YE, Cheng CM, Yang CH, Tsai SJ, Lu CL, Tsai CF. Association between gastrointestinal symptoms and depression among older adults in Taiwan: A cross-sectional study. J Chin Med Assoc. 2021;84(3):331–5. Sharpe L, McDonald S, Correia H, Raue PJ, Meade T, Nicholas M, Arean P. Pain severity predicts depressive symptoms over and above individual illnesses and multimorbidity in older adults. BMC Psychiatry. 2017;17(1):166. Ye X, Wang X. Associations of multimorbidity with body pain, sleep duration, and depression among middle-aged and older adults in China. Health Qual Life Outcomes. 2024;22(1):23. Kroenke K, Wu J, Bair MJ, Krebs EE, Damush TM, Tu W. Reciprocal relationship between pain and depression: a 12-month longitudinal analysis in primary care. J Pain. 2011;12(9):964–73. Ho HE, Yeh CJ, Cheng-Chung Wei J, Chu WM, Lee MC. Association between multimorbidity patterns and incident depression among older adults in Taiwan: the role of social participation. BMC Geriatr. 2023;23(1):177. McClain AC, Xiao RS, Tucker KL, Falcón LM, Mattei J. Depressive symptoms and allostatic load have a bidirectional association among Puerto Rican older adults. Psychol Med. 2022;52(14):3073–85. Griffin SC, Ravyts SG, Bourchtein E, Ulmer CS, Leggett MK, Dzierzewski JM, Calhoun PS. Sleep disturbance and pain in U.S. adults over 50: evidence for reciprocal, longitudinal effects. Sleep Med. 2021;86:32–9. Lingying W, Hong Z, Hongxiu C, Ziyi H, Mei F, Menglin T, Xiuying H. Association of body mass index with disability in activities of daily living in older adults: a systematic review of the literature based on longitudinal data. BMC Public Health. 2025;25(1):6. Li X, Feng H, Chen Q. Social participation patterns and associations with subsequent cognitive function in older adults with cognitive impairment: a latent class analysis. Front Med (Lausanne). 2025;12:1493359. Yin F, Fan X, Chen J, Wu X. Leisure activity patterns and health vulnerability among older adults in China: a nationwide analysis of urban-rural differences based on CHARLS data. BMC Geriatr 2026, 26(1). Jia H, Lubetkin EI. Association between self-reported body mass index and active life expectancy in a large community-dwelling sample of older U.S. adults. BMC Geriatr. 2022;22(1):310. Pan H, Hao Y. Body Mass Index and multidimensional health in Chinese older adults: a moderated mediation analysis of urban-rural residence, physical activity, and depression. Front Public Health. 2025;13:1650975. Marengoni A, Tazzeo C, Calderón-Larrañaga A, Roso-Llorach A, Onder G, Zucchelli A, Rizzuto D, Vetrano DL. Multimorbidity Patterns and 6-Year Risk of Institutionalization in Older Persons: The Role of Social Formal and Informal Care. J Am Med Dir Assoc. 2021;22(10):2184–e21892181. Barrio-Cortes J, Castaño-Reguillo A, Benito-Sánchez B, Beca-Martínez MT, Ruiz-Zaldibar C. Utilization of Primary Healthcare Services in Patients with Multimorbidity According to Their Risk Level by Adjusted Morbidity Groups: A Cross-Sectional Study in Chamartín District (Madrid). Healthc (Basel) 2024, 12(2). Smith SM, Wallace E, Clyne B, Boland F, Fortin M. Interventions for improving outcomes in patients with multimorbidity in primary care and community setting: a systematic review. Syst Rev. 2021;10(1):271. Additional Declarations No competing interests reported. Cite Share Download PDF Status: Under Review Version 1 posted Editorial decision: Revision requested 04 May, 2026 Reviews received at journal 24 Apr, 2026 Reviews received at journal 17 Apr, 2026 Reviewers agreed at journal 03 Apr, 2026 Reviewers agreed at journal 30 Mar, 2026 Reviewers agreed at journal 29 Mar, 2026 Reviewers invited by journal 29 Mar, 2026 Editor invited by journal 16 Mar, 2026 Editor assigned by journal 13 Mar, 2026 Submission checks completed at journal 13 Mar, 2026 First submitted to journal 10 Mar, 2026 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-9085989","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":615371312,"identity":"71092569-2512-4458-8599-3cdad03d689b","order_by":0,"name":"Kaige Gao","email":"","orcid":"","institution":"Xinjiang Medical University","correspondingAuthor":false,"prefix":"","firstName":"Kaige","middleName":"","lastName":"Gao","suffix":""},{"id":615371313,"identity":"968bc605-f4e7-41c7-8492-5b9253fecd17","order_by":1,"name":"Yanjie Zhao","email":"","orcid":"","institution":"Xinjiang Medical University","correspondingAuthor":false,"prefix":"","firstName":"Yanjie","middleName":"","lastName":"Zhao","suffix":""},{"id":615371314,"identity":"6abebc86-475c-48de-aa77-713a135088d6","order_by":2,"name":"Xuehong Zhang","email":"","orcid":"","institution":"Xingfu Road Community Health Service Center","correspondingAuthor":false,"prefix":"","firstName":"Xuehong","middleName":"","lastName":"Zhang","suffix":""},{"id":615371315,"identity":"cc59301e-60da-4921-8e92-ae416ac79d4c","order_by":3,"name":"Kanbao Su","email":"","orcid":"","institution":"Xinjiang Neutrino Artificial Intelligence Technology Co., Ltd.","correspondingAuthor":false,"prefix":"","firstName":"Kanbao","middleName":"","lastName":"Su","suffix":""},{"id":615371318,"identity":"bedc142d-b0a9-4612-bfc5-1aaf34007273","order_by":4,"name":"Jia Jia","email":"","orcid":"","institution":"Xinjiang Neutrino Artificial Intelligence Technology Co., Ltd.","correspondingAuthor":false,"prefix":"","firstName":"Jia","middleName":"","lastName":"Jia","suffix":""},{"id":615371328,"identity":"72cd9ae8-8f25-471e-ae30-8a3f40fd3536","order_by":5,"name":"Yuezhen Xu","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAv0lEQVRIiWNgGAWjYDACdh42hg8GEjz8zMyHHxCnhZmHjXFGhY2cZDtbmgHRWph5zqQZG5znUZAgSoe8M++xB7xthxM3H+ZhMGCosYkmqMXwMF+6gSRQy7bDvAceMBxLy20gqKWZx0zCEKyFL8GAseEwkVoSQQ5r5jGQIEqLPDNQywGQ95mJ1WLAzJcm2QAMZInDwEBOIMYv8u29x6T/gKKy//DhBx9qbIiw5QAyL4GQcrAtBA0dBaNgFIyCUQAAcZk8FkorXlQAAAAASUVORK5CYII=","orcid":"","institution":"Xinjiang Medical University","correspondingAuthor":true,"prefix":"","firstName":"Yuezhen","middleName":"","lastName":"Xu","suffix":""}],"badges":[],"createdAt":"2026-03-10 15:53:32","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-9085989/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-9085989/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":106030533,"identity":"24e499f6-a73e-47d5-bdeb-42ccf5a22cb7","added_by":"auto","created_at":"2026-04-02 15:16:54","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":164563,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eConditional probabilities of chronic conditions by latent class (K = 4)\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNote: Darker shading indicates a higher model-estimated conditional probability of the corresponding chronic condition within each latent class.\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-9085989/v1/bccf58a7d7037806b299dd1b.png"},{"id":106094390,"identity":"a490afe0-9f39-43ef-a453-898d3f9bee7a","added_by":"auto","created_at":"2026-04-03 11:42:24","extension":"jpg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":82581,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003ePredicted trajectories of depressive symptoms (CES-D scores) across multimorbidity patterns among middle-aged and older adults, 2013–2018\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNote: Figure 2 illustrates the longitudinal trajectories of CES-D scores across the four multimorbidity patterns from 2013 to 2018. Scores were estimated using linear mixed-effects models (LMMs) with 20 multiple imputations. Lines represent predicted means; shaded areas indicate 95% CIs pooled via Rubin\u003cstrong\u003e'\u003c/strong\u003es rules. The model adjusted for all baseline (2011) covariates, including baseline CES-D, demographics, and health behaviours. Follow-up year 0 corresponds to 2013.\u003c/p\u003e","description":"","filename":"2.jpg","url":"https://assets-eu.researchsquare.com/files/rs-9085989/v1/2a1e40b5b8dacc89b2327c03.jpg"},{"id":106959653,"identity":"0326d333-25b8-42d4-b6a1-3bec8f5add80","added_by":"auto","created_at":"2026-04-15 09:12:56","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1740244,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9085989/v1/4b0b6912-1e8c-40d0-8cad-ad1c716d1461.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Latent multimorbidity patterns and their longitudinal associations with depressive symptom trajectories and ADL limitations among middle-aged and older adults in China: a longitudinal analysis of the CHARLS","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003ePopulation aging is accelerating worldwide, alongside a growing burden of chronic disease among middle-aged and older adults. Multimorbidity has become an increasing challenge for healthcare systems, particularly for care coordination, long-term care planning, and resource allocation[\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e, \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e]. In addition to disease accumulation, multimorbidity is often accompanied by greater symptom burden, polypharmacy, and a higher risk of drug\u0026ndash;drug interactions, and has been associated with increased healthcare use, functional decline, and poorer health-related quality of life[\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e]. In healthy aging and chronic disease management, identifying high-risk groups and understanding how psychological and functional outcomes change over time may help improve risk stratification, intervention planning, and resource allocation[\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eMultimorbidity is not a homogeneous construct. In epidemiological research, it is commonly defined using a threshold approach (e.g., \u0026ge;\u0026thinsp;2 chronic conditions) or simple disease counts. Although these approaches are useful for quantifying burden, they may overlook differences in how chronic conditions cluster and in the clinical implications of specific combinations[\u003cspan additionalcitationids=\"CR5\" citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]. Evidence suggests that chronic conditions tend to cluster into relatively stable patterns, such as cardiometabolic or respiratory\u0026ndash;musculoskeletal groupings[\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]. These patterns may reflect different underlying mechanisms and risk profiles and may therefore be associated with different trajectories of mental and functional outcomes. From a public health perspective, examining specific multimorbidity patterns may be more informative than relying on disease counts alone when assessing patient-centered outcomes.\u003c/p\u003e \u003cp\u003eLatent class analysis (LCA) is a data-driven method for identifying such patterns. By modelling multiple binary disease indicators simultaneously, LCA classifies individuals into latent subgroups with similar co-occurrence structures and provides probabilistic class assignments, allowing the identification of interpretable multimorbidity patterns[\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e]. Although the use of LCA in multimorbidity research has increased in recent years, most studies remain cross-sectional. In longitudinal research, analyses often focus on a single outcome or a single time point, and less is known about whether baseline multimorbidity patterns are associated with different changes in psychological and functional outcomes over time[\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e]. Evidence linking baseline multimorbidity patterns to subsequent outcome trajectories therefore remains limited.\u003c/p\u003e \u003cp\u003eThis question is particularly relevant in China, where population aging is occurring rapidly. National surveys indicate that multimorbidity among adults aged 60 years and older is common, with reported prevalence estimates of approximately 46\u0026ndash;57%, and the burden continues to increase[\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e, \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e]. These patterns reflect not only the epidemiological transition but also persistent urban\u0026ndash;rural disparities and inequalities in healthcare access. Despite the scale of the burden, longitudinal evidence on how distinct multimorbidity patterns relate to mental and functional outcomes in the Chinese population remains limited [\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eDepressive symptoms and limitations in activities of daily living (ADL) are two indicators of aging-related vulnerability. Depressive symptoms are commonly measured using standardized instruments and are associated with healthcare use and mortality risk. ADL limitation reflects impairment in basic self-care and is a predictor of disability progression and long-term care needs[\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e]. Previous longitudinal studies suggest that psychological distress and functional decline may influence one another over time, but it remains unclear whether different multimorbidity structures are associated with different trajectories of these outcomes in the Chinese context[\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e, \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eUsing nationally representative data from the China Health and Retirement Longitudinal Study (CHARLS), this study aimed to identify latent multimorbidity patterns at baseline (2011) and examine their longitudinal associations with subsequent changes in depressive symptoms and ADL limitation from 2013 to 2018. Specifically, we aimed to: (1) use LCA to identify latent classes based on baseline chronic disease indicators; (2) assess differences in depressive symptom levels and the risk of ADL limitation across classes during follow-up; and (3) examine whether rates of change in these outcomes differed across classes through class-by-time interaction terms.\u003c/p\u003e"},{"header":"2. Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003e2.1 Study design and sample\u003c/h2\u003e \u003cp\u003eThis prospective observational study examined whether baseline multimorbidity patterns were associated with subsequent longitudinal outcomes. Data came from the China Health and Retirement Longitudinal Study (CHARLS), a nationally representative panel survey of Chinese adults aged 45 years and older conducted by the National School of Development at Peking University[\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e]. Across repeated survey waves, CHARLS collects information on demographic characteristics, socioeconomic status, health conditions, and functional status. For the present analysis, the 2011 wave served as baseline, and follow-up information was taken from the 2013, 2015, and 2018 waves. The dataset is publicly available upon approved application.\u003c/p\u003e \u003cp\u003eAt baseline, CHARLS included 17,708 respondents. Because latent class membership could not be estimated for participants with missing data on all 12 chronic disease indicators, these individuals were excluded. This left 17,091 participants with non-missing most-likely latent class assignments for the latent class analysis (LCA) (Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e3\u003c/span\u003e). We further excluded respondents with implausible baseline age values and those who had no valid outcome observations during follow-up from 2013 to 2018. After these exclusions, the final longitudinal cohort comprised 15,879 individuals. Among them, 14,725 contributed to the depressive symptom models and 13,673 to the ADL limitation models. The difference in analytic sample size mainly reflected outcome-specific missingness and attrition across waves.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eEstimated and Most Likely Latent Class Proportions (N\u0026thinsp;=\u0026thinsp;17,091)\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"4\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLatent Class\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eModel-estimated Proportion\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eMost-likely Class Proportion\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eClinical Interpretation\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eClass 4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e69.11%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e77.10%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eRelatively healthy\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eClass 1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e13.82%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e10.58%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eCardiometabolic\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eClass 3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e14.12%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e10.01%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eMusculoskeletal\u0026ndash;digestive\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eClass 2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e2.95%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e2.32%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eRespiratory-dominant\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"4\"\u003eNote: N denotes the number of participants with non-missing most-likely class assignments obtained from the baseline latent class analysis (LCA). Model parameters in the LCA were estimated using full information maximum likelihood (FIML) to account for partial missing data in the chronic disease indicators.\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec4\" class=\"Section2\"\u003e \u003ch2\u003e2.2 Variable measurement and definitions\u003c/h2\u003e \u003cdiv id=\"Sec5\" class=\"Section3\"\u003e \u003ch2\u003e2.2.1 Exposure: Chronic disease assessment and latent class identification\u003c/h2\u003e \u003cp\u003eBaseline multimorbidity was defined using 12 chronic conditions assessed in 2011 on the basis of self-reported physician diagnoses or medical history. The conditions were hypertension, dyslipidemia, diabetes, malignant tumour, chronic lung disease, liver disease, heart disease, stroke, kidney disease, digestive disease, arthritis, and asthma. In large epidemiological surveys, self-reported physician diagnoses are commonly used. Studies based on CHARLS have shown moderate agreement and high specificity when these reports are compared with objective biomedical measurements, supporting their use in population-based research [\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e, \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eTo identify multimorbidity patterns, we conducted LCA in Mplus version 8.3 using the 12 binary chronic disease indicators. Models with two to six classes were estimated. Selection of the final model considered several criteria: the Akaike Information Criterion (AIC), Bayesian Information Criterion (BIC), sample-size adjusted BIC (aBIC), entropy, the Lo\u0026ndash;Mendell\u0026ndash;Rubin adjusted likelihood ratio test (LMR), the bootstrap likelihood ratio test (BLRT), class size, and clinical interpretability. Within the LCA framework, missing data on individual disease indicators were handled using full information maximum likelihood (FIML), so all available responses could contribute to estimation. Participants with missing values on all 12 indicators were excluded because class membership could not be estimated.\u003c/p\u003e \u003cp\u003eOnce the final model had been selected, participants were assigned to latent classes using the most-likely class approach based on maximum posterior probability (MaxPP). Class membership was then entered into subsequent regression analyses as a categorical exposure variable, with the relatively healthy class as the reference group. Posterior probabilities were retained to evaluate classification quality. To assess the possible influence of classification uncertainty, we repeated key analyses in subsamples restricted to participants with higher assignment certainty according to prespecified MaxPP thresholds.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec6\" class=\"Section3\"\u003e \u003ch2\u003e2.2.2 Outcomes\u003c/h2\u003e \u003cp\u003eDepressive symptoms were measured with the 10-item Center for Epidemiologic Studies Depression Scale (CES-D-10). This instrument has been validated in Chinese populations and has shown acceptable reliability and construct validity[\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e]. Scores range from 0 to 30, with higher values indicating more severe depressive symptoms. Before summation, the two positively worded items were reverse-coded.\u003c/p\u003e \u003cp\u003eFor the primary longitudinal analyses, CES-D-10 total score was modelled as a continuous outcome. A conventional cut-off of \u0026ge;\u0026thinsp;10 was used for descriptive purposes to indicate elevated depressive symptoms, but it was not adopted in the primary modelling strategy unless otherwise specified.\u003c/p\u003e \u003cp\u003eADL limitation was defined as reporting difficulty or requiring assistance in at least one basic activity of daily living, namely dressing, bathing, eating, getting in or out of bed, toileting, and continence. At each follow-up wave, this information was used to construct a binary variable (any limitation vs. none). This variable was then used to examine change in the risk of ADL limitation over time.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec7\" class=\"Section3\"\u003e \u003ch2\u003e2.2.3 Covariates\u003c/h2\u003e \u003cp\u003eAll covariates were measured at baseline in 2011 and treated as time-invariant in the longitudinal analyses. They included age, sex, educational attainment, marital status, urban\u0026ndash;rural residence, body mass index (BMI), and smoking status.\u003c/p\u003e \u003cp\u003eBecause baseline levels of the outcomes may influence subsequent change, the models also adjusted for the corresponding baseline outcome measure. Specifically, the depressive symptom models included the 2011 CES-D score, whereas the ADL models included baseline ADL limitation status. This adjustment was intended to reduce confounding by initial health status.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003e2.3 Statistical analysis\u003c/h2\u003e \u003cp\u003eTo analyse the longitudinal associations between baseline multimorbidity patterns and subsequent outcomes, we constructed a person-wave dataset in long format and fitted mixed-effects models. Depressive symptoms (CES-D-10 total score) were analysed using linear mixed-effects models (LMMs), whereas ADL limitation was analysed using logistic generalized linear mixed-effects models (GLMMs). In all models, a random intercept was included to account for between-individual differences in baseline outcome levels. The main parameter of interest was the interaction between latent class membership and time, which tested whether rates of change in the outcomes differed across multimorbidity patterns.\u003c/p\u003e \u003cp\u003eTime was parameterized as years since the first post-baseline follow-up wave (2013) and coded 0, 2, and 5 for the 2013, 2015, and 2018 waves, respectively. Exposure, covariates, and baseline outcome levels were all defined from the 2011 wave, which was therefore not included in the trajectory models. Model adjustment covered baseline demographic characteristics, socioeconomic factors, health behaviours, and the corresponding baseline outcome measure.\u003c/p\u003e \u003cp\u003eMissing baseline covariate data were addressed using multiple imputation by chained equations (MICE; m\u0026thinsp;=\u0026thinsp;20), after which estimates were pooled according to Rubin\u0026rsquo;s rules [\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e, \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e]. The LCA was performed before imputation based on the baseline chronic disease indicators, and missingness in those indicators was handled in Mplus using FIML. For that reason, the LCA indicators themselves were not imputed. Missing follow-up outcome data were handled under the maximum likelihood framework of the mixed-effects models, assuming missing at random (MAR). Sensitivity analyses considered alternative parameterizations of time, random-slope models, and analyses restricted to outcome-free subsamples at baseline. In addition, exploratory mediation analysis was conducted within a counterfactual framework to estimate the statistical indirect effect of pain on subsequent depressive symptoms.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec9\" class=\"Section2\"\u003e \u003ch2\u003e2.4 Robustness and classification certainty analyses\u003c/h2\u003e \u003cp\u003eSeveral sensitivity analyses were performed to examine whether the primary findings were robust to alternative model specifications and sample restrictions.\u003c/p\u003e \u003cp\u003eWe first considered alternative ways of modelling time. To allow for potential non-linear trends, the continuous time variable was replaced with categorical wave indicators, using 2013 as the reference. In the depressive symptom models, we also fitted random-slope models for time to account for between-individual variation in rates of change.\u003c/p\u003e \u003cp\u003eWe next repeated the primary analyses in restricted subsamples to reduce the possibility that baseline outcome status influenced subsequent associations. For depressive symptoms, analyses were limited to participants without elevated depressive symptoms at baseline (CES-D\u0026thinsp;\u0026lt;\u0026thinsp;10). For ADL limitation, analyses were restricted to those without baseline ADL limitation. These models were used to examine whether the observed associations were similar among participants free of the corresponding outcome at study entry.\u003c/p\u003e \u003cp\u003eClassification certainty was examined in a further set of analyses. The primary models were based on most-likely class assignment. To assess whether classification uncertainty materially affected the estimates, we re-estimated the models after restricting the sample to participants with higher assignment certainty, defined by maximum posterior probability (MaxPP) thresholds of \u0026ge;\u0026thinsp;0.70 and \u0026ge;\u0026thinsp;0.80.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec10\" class=\"Section2\"\u003e \u003ch2\u003e2.5 Mechanistic exploration: mediation analysis\u003c/h2\u003e \u003cp\u003eTo explore a possible pathway underlying the association between the musculoskeletal\u0026ndash;digestive pattern (Class 3) and subsequent depressive symptoms, we conducted an exploratory mediation analysis within a counterfactual framework following Imai et al[\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e, \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e]. Latent class membership in 2011 was treated as the exposure, pain indicators measured in 2013 as the mediator, and depressive symptom scores measured in 2015 and 2018 as the outcome. Pain was examined in three forms: any pain, pain severity, and number of pain sites. The primary comparison was between Class 3 and the relatively healthy class (Class 4).\u003c/p\u003e \u003cp\u003eMediator models were specified according to the type of mediator, using logistic regression for binary pain indicators and linear regression for continuous or count-based pain indicators. Outcome models were estimated using linear regression adjusted for baseline depressive symptoms, baseline demographic and health-behaviour covariates, and survey wave. The average causal mediation effect (ACME) and average direct effect (ADE) were estimated.\u003c/p\u003e \u003cp\u003eFor the multiply imputed datasets (MICE, m\u0026thinsp;=\u0026thinsp;20), mediation analyses were conducted separately within each imputed dataset and pooled according to Rubin\u0026rsquo;s rules. Given the observational design and the assumptions required for mediation analysis, these results were interpreted as exploratory evidence relevant to potential mechanisms rather than as confirmation of a causal pathway.\u003c/p\u003e \u003c/div\u003e"},{"header":"3. Results","content":"\u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003e3.1 Identification of latent classes\u003c/h2\u003e \u003cp\u003eUsing the 12 binary chronic disease indicators measured at baseline in 2011, we fitted latent class models with two to six classes and compared their fit (Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e1\u003c/span\u003e). Model selection considered statistical fit indices, likelihood ratio tests, entropy, class size, and clinical interpretability. On balance, the four-class model was selected for the primary analyses.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eModel Fit Indices for Latent Class Analysis (K\u0026thinsp;=\u0026thinsp;2\u0026ndash;6)\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"8\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c8\" colnum=\"8\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eClasses (K)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eLL\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eAIC\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eBIC\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eaBIC\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eEntropy\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c7\"\u003e \u003cp\u003eLMR P-value\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c8\"\u003e \u003cp\u003eBLRT P-value\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e2 Classes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e-60820.794\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e121,691.59\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e121,885.25\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e121,805.80\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.543\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e3 Classes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e-60113.081\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e120,302.16\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e120,596.52\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e120,475.76\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.636\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e4 Classes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e-59816.962\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e119,735.93\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e120,130.99\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e119,968.91\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.702\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e5 Classes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e-59660.225\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e119,448.45\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e119,944.21\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e119,740.83\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.63\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e6 Classes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e-59619.025\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e119,392.05\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e119,988.52\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e119,743.82\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.647\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.046\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"8\"\u003eNote: LL\u0026thinsp;=\u0026thinsp;log-likelihood; AIC\u0026thinsp;=\u0026thinsp;Akaike Information Criterion; BIC\u0026thinsp;=\u0026thinsp;Bayesian Information Criterion; aBIC\u0026thinsp;=\u0026thinsp;sample-size adjusted BIC. The four-class solution was selected as the final latent class model.\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eThe four-class model had an entropy of 0.702, higher than that of the three-class model (0.636) and the five-class model (0.630) (Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e1\u003c/span\u003e). Although the five-class model showed a slightly lower BIC than the four-class model, the BIC increased again in the six-class model. In addition, models with more classes tended to yield smaller class sizes and lower classification quality. Taken together, these results supported the four-class solution as the most interpretable and parsimonious model. The LMR and BLRT also indicated improved fit up to the four-class model (Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e1\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eConditional probabilities from the four-class model suggested four distinct multimorbidity patterns (Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e2\u003c/span\u003e and Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). Class 1 was labelled cardiometabolic, characterized by higher probabilities of hypertension, dyslipidemia, and diabetes. Class 2 was labelled respiratory-dominant, with chronic lung disease and asthma as the main conditions. Class 3 was labelled musculoskeletal\u0026ndash;digestive, marked by arthritis and stomach or digestive disease. Class 4 was labelled relatively healthy, with low probabilities across all conditions. In the five- and six-class models, additional classes mainly reflected further subdivision of these patterns or smaller subgroups with limited added clinical interpretability.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eItem-Response Probabilities for the Four-Class Model\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eChronic Conditions\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eCardiometabolic\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eRespiratory-dominant\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eMusculoskeletal\u0026ndash;digestive\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eRelatively healthy\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ehypertension\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.784\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.277\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.288\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.133\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003edyslipidemia\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.411\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.103\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.111\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.027\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ediabetes mellitus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.261\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.067\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.055\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.018\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eheart disease\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.363\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.294\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.284\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.033\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003estroke\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.109\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.034\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.026\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.006\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003echronic lung disease\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.071\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.233\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.045\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003easthma\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.023\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.702\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.05\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.009\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003earthritis\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.313\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.443\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.798\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.237\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003estomach or digestive disease\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.148\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.293\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.62\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.154\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ekidney disease\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.078\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.101\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.221\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.028\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eliver disease\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.045\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.056\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.133\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.018\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ecancer\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.017\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.016\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.008\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"5\"\u003eNote: Values represent the item\u0026ndash;response (conditional) probabilities estimated from the four-class latent class model. Probabilities indicate the likelihood of reporting the presence of each chronic condition (coded as 1) given membership in a specific latent class. Higher values indicate greater concentration of the condition within that class. These probabilities are model estimates and do not represent crude prevalence in the overall sample.\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eAs shown in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e3\u003c/span\u003e, the model-estimated class proportions and the most-likely class assignments were broadly similar. Because some degree of classification uncertainty remained, we conducted additional analyses using maximum posterior probability (MaxPP) thresholds of \u0026ge;\u0026thinsp;0.70 and \u0026ge;\u0026thinsp;0.80. Key class-by-time interaction terms were then re-estimated in subsamples with higher assignment certainty to examine whether the direction and magnitude of the estimates changed materially.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003e3.2 Baseline characteristics across multimorbidity patterns\u003c/h2\u003e \u003cp\u003eAmong the 15,879 participants included in the longitudinal cohort, several baseline characteristics differed across the four latent classes (Table\u0026nbsp;\u003cspan refid=\"Tab4\" class=\"InternalRef\"\u003e4\u003c/span\u003e). Age, sex, and marital status did not differ significantly across classes (all P\u0026thinsp;\u0026gt;\u0026thinsp;0.05). In contrast, educational attainment, urban\u0026ndash;rural residence, and BMI showed significant between-class differences (all P\u0026thinsp;\u0026lt;\u0026thinsp;0.01). Participants in the cardiometabolic class (Class 1) had the highest mean BMI.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab4\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 4\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eBaseline characteristics across latent multimorbidity classes among participants included in the longitudinal cohort (N\u0026thinsp;=\u0026thinsp;15,879)\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"6\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCharacteristic\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eRelatively healthy\u003c/p\u003e \u003cp\u003e(n\u0026thinsp;=\u0026thinsp;12 ,249)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eCardiometabolic\u003c/p\u003e \u003cp\u003e(n\u0026thinsp;=\u0026thinsp;1 ,669)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eRespiratory-dominant (n\u0026thinsp;=\u0026thinsp;362)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eMusculoskeletal\u0026ndash;digestive\u003c/p\u003e \u003cp\u003e(n\u0026thinsp;=\u0026thinsp;1 ,599)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eP value\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAge (mean (SD))\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e58.82 (9.66)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e58.49 (9.20)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e58.73 (9.16)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e58.94 (9.73)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.554\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFemale, n (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e6315 (51.6)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e865 (51.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e172 (47.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e816 (51.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.467\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eEducation level, %\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.006\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNo formal education\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e3346 (27.4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e418 (25.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e104 (28.7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e450 (28.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePrimary school\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e4868 (39.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e625 (37.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e131 (36.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e664 (41.6)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMiddle school\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e2496 (20.4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e391 (23.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e83 (22.9)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e311 (19.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHigh school and above\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e1524 (12.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e232 (13.9)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e44 (12.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e173 (10.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eResidence (Urban), %\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e5641 (46.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e809 (48.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e172 (47.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e734 (46.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMarital status (Married), %\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e10717 (87.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1484 (88.9)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e317 (87.6)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e1388 (86.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.301\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCurrent smoking, %\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e3494 (29.6)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e464 (28.7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e108 (30.9)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e445 (28.7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.698\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBMI (mean (SD))\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e23.48 (3.93)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e24.21 (4.37)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e23.35 (3.61)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e23.45 (3.91)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBaseline CES-D-10 score, mean (SD)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e8.45 (6.34)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e8.17 (6.53)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e8.46 (6.55)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e8.75 (6.22)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.106\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNo ADL limitation, %\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e6408 (76.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e894 (77.7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e202 (78.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e886 (75.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.4\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"6\"\u003eNote: Continuous variables are presented as mean (standard deviation, SD), and categorical variables as number (percentage, %). Baseline depressive symptoms were assessed using the CES-D-10 scale (range: 0\u0026ndash;30), with higher scores indicating more severe symptoms. Limitation in activities of daily living (ADL) was defined as dependence on or need for assistance with at least one basic ADL task. P-values for continuous variables were derived from one-way analysis of variance (ANOVA), and P-values for categorical variables from chi-square tests. The total sample entering the longitudinal analyses was N\u0026thinsp;=\u0026thinsp;15,879. Sample sizes varied slightly across outcome models due to missing data; details are provided in the Methods section.\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eThe crude distribution of individual chronic conditions across classes was generally consistent with the conditional probabilities estimated from the LCA model (Table\u0026nbsp;\u003cspan refid=\"Tab4\" class=\"InternalRef\"\u003e4\u003c/span\u003e). Baseline CES-D-10 scores and the prevalence of ADL limitation were similar across classes, with no statistically significant differences observed (both P\u0026thinsp;\u0026gt;\u0026thinsp;0.05). These baseline characteristics provide context for the subsequent longitudinal analyses, in which all models were adjusted for baseline covariates and the corresponding baseline outcome measure.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec14\" class=\"Section2\"\u003e \u003ch2\u003e3.3 Longitudinal associations between multimorbidity patterns and depressive symptom trajectories\u003c/h2\u003e \u003cp\u003eIn linear mixed-effects models pooled across 20 imputed datasets, CES-D-10 scores increased over follow-up from 2013 to 2018 (β\u0026thinsp;=\u0026thinsp;0.200 per year, 95% CI: 0.171\u0026ndash;0.229; P\u0026thinsp;\u0026lt;\u0026thinsp;0.001) (Table\u0026nbsp;\u003cspan refid=\"Tab5\" class=\"InternalRef\"\u003e5\u003c/span\u003e). Adjusted predicted trajectories are shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e. Compared with the relatively healthy class (Class 4), the musculoskeletal\u0026ndash;digestive class (Class 3) was the only class with a significant class-by-time interaction (β\u0026thinsp;=\u0026thinsp;0.099, 95% CI: 0.016\u0026ndash;0.183; P\u0026thinsp;=\u0026thinsp;0.020), corresponding to a steeper increase in depressive symptom scores over time. Baseline depressive symptom scores did not differ significantly between Class 3 and Class 4 (P\u0026thinsp;=\u0026thinsp;0.538). For the cardiometabolic class (Class 1), the class-by-time interaction was positive but did not reach statistical significance (P\u0026thinsp;=\u0026thinsp;0.083).\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab5\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 5\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eLongitudinal Associations Between Multimorbidity Patterns and Depressive Symptom Trajectories (CES-D Scores) Among Middle-Aged and Older Adults, 2013\u0026ndash;2018\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"3\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eVariable\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eβ (95% CI)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eP value\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBaseline level, vs. Class 4 (Relatively healthy)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCardiometabolic (Class 1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e-0.142 (-0.459, 0.176)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.382\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRespiratory-dominant (Class 2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.128 (-0.499, 0.754)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.689\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMusculoskeletal\u0026ndash;digestive (Class 3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.101 (-0.221, 0.423)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.538\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTime effect (per year)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.200 (0.171, 0.229)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eInteraction terms: differences in rate of change\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCardiometabolic \u0026times; Time\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.074 (-0.010, 0.157)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.083\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRespiratory-dominant \u0026times; Time\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.037 (-0.127, 0.201)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.658\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMusculoskeletal\u0026ndash;digestive \u0026times; Time\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.099 (0.016, 0.183)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.020\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"3\"\u003eNote: CES-D\u0026thinsp;=\u0026thinsp;Center for Epidemiologic Studies Depression Scale; CI\u0026thinsp;=\u0026thinsp;confidence interval. All estimates were obtained from linear mixed-effects models (LMMs) and pooled across 20 multiply imputed datasets using Rubin\u0026rsquo;s rules. Interaction terms represent the estimated differences in the rate of change in depressive symptoms over time for each multimorbidity pattern relative to the reference group. Statistical significance was defined as P\u0026thinsp;\u0026lt;\u0026thinsp;0.05.\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec15\" class=\"Section2\"\u003e \u003ch2\u003e3.4 Longitudinal associations between multimorbidity patterns and risk of ADL limitation\u003c/h2\u003e \u003cp\u003eIn logistic generalized linear mixed-effects models pooled across 20 imputed datasets, the odds of ADL limitation increased over time (OR\u0026thinsp;=\u0026thinsp;1.083 per year, 95% CI: 1.064\u0026ndash;1.103; P\u0026thinsp;\u0026lt;\u0026thinsp;0.001) (Table\u0026nbsp;\u003cspan refid=\"Tab6\" class=\"InternalRef\"\u003e6\u003c/span\u003e). After adjustment for baseline ADL limitation status and covariates, the musculoskeletal\u0026ndash;digestive class (Class 3: OR\u0026thinsp;=\u0026thinsp;1.141, 95% CI: 0.959\u0026ndash;1.358; P\u0026thinsp;=\u0026thinsp;0.138) and the cardiometabolic class (Class 1: OR\u0026thinsp;=\u0026thinsp;1.136, 95% CI: 0.950\u0026ndash;1.359; P\u0026thinsp;=\u0026thinsp;0.162) showed higher point estimates than the relatively healthy class, although these differences were not statistically significant.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab6\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 6\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eLongitudinal Associations Between Multimorbidity Patterns and Risk of ADL Limitation Among Middle-Aged and Older Adults, 2013\u0026ndash;2018\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"3\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eVariable\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eOR (95% CI)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eP value\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBaseline level, vs. Class 4 (Relatively healthy)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCardiometabolic (Class 1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e1.136 (0.950, 1.359)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.162\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRespiratory-dominant (Class 2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.949 (0.659, 1.367)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.780\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMusculoskeletal\u0026ndash;digestive (Class 3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e1.141 (0.959, 1.357)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.138\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTime effect (per year)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e1.083 (1.064, 1.103)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eInteraction terms: differences in rate of change\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCardiometabolic \u0026times; Time\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e1.029 (0.978, 1.082)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.268\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRespiratory-dominant \u0026times; Time\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e1.029 (0.930, 1.140)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.576\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMusculoskeletal\u0026ndash;digestive \u0026times; Time\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.984 (0.936, 1.033)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.515\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"3\"\u003eNote: ADL\u0026thinsp;=\u0026thinsp;activities of daily living; OR\u0026thinsp;=\u0026thinsp;odds ratio; CI\u0026thinsp;=\u0026thinsp;confidence interval. Estimates were derived from generalized linear mixed-effects models (GLMMs) and pooled across 20 multiply imputed datasets using Rubin\u0026rsquo;s rules. Class 4 (relatively healthy) served as the reference group. Interaction terms represent differences in the rate of change in ADL limitation risk over time between each multimorbidity pattern and the reference group. Statistical significance was defined as P\u0026thinsp;\u0026lt;\u0026thinsp;0.05.\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eNo class-by-time interaction terms reached statistical significance (all P\u0026thinsp;\u0026gt;\u0026thinsp;0.05), suggesting that the rate of change in ADL limitation risk did not differ across multimorbidity classes under the specified model.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec16\" class=\"Section2\"\u003e \u003ch2\u003e3.5 Robustness and sensitivity analyses\u003c/h2\u003e \u003cp\u003eSensitivity analyses were conducted to examine the robustness of the primary findings under alternative model specifications and sample restrictions (Table\u0026nbsp;\u003cspan refid=\"Tab7\" class=\"InternalRef\"\u003e7\u003c/span\u003e). Among participants without elevated depressive symptoms at baseline, the class-by-time interaction for the musculoskeletal\u0026ndash;digestive class (Class 3) was similar in magnitude to that in the primary analysis (β\u0026thinsp;=\u0026thinsp;0.095 vs. 0.099), although the estimate was less precise in the smaller sample (P\u0026thinsp;=\u0026thinsp;0.059). In random-slope models that allowed individual-specific rates of change, the Class 3 interaction remained similar to the primary estimate (β\u0026thinsp;=\u0026thinsp;0.096; P\u0026thinsp;=\u0026thinsp;0.028). When time was parameterized using wave indicators, depressive symptoms increased across follow-up waves, and the difference between Class 3 and the relatively healthy class was larger at the 2018 wave (β for Class 3 vs. Class 4 at 2018 relative to 2013\u0026thinsp;=\u0026thinsp;0.496; P\u0026thinsp;=\u0026thinsp;0.020), consistent with the positive class-by-time interaction observed in the primary model.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab7\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 7\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eSummary of Robustness Analyses for the Association Between the Musculoskeletal\u0026ndash;Digestive Multimorbidity Pattern and Health Outcomes\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRobustness analysis model\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eDepressive symptoms (DEP) β (95% CI)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eP value\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eADL limitation OR (95% CI)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eP value\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePrimary analysis model\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.099 (0.016, 0.183)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.020\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.984 (0.936, 1.033)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.515\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNonlinear time specification\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.496 (0.077, 0.915)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.020\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.933 (0.725, 1.200)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.589\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRandom-slope model\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.096 (0.010, 0.182)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.028\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.984 (0.937, 1.034)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.519\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBaseline outcome exclusion\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.095 (-0.004, 0.193)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.059\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.983 (0.925, 1.043)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.564\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHigh classification certainty sample\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.102 (-0.030, 0.234)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.131\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.994 (0.921, 1.072)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.874\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"5\"\u003eNote: DEP\u0026thinsp;=\u0026thinsp;depressive symptoms; ADL\u0026thinsp;=\u0026thinsp;activities of daily living. This table summarizes the interaction coefficients between the musculoskeletal\u0026ndash;digestive multimorbidity pattern (Class 3) and time (or survey wave) across multiple robustness analyses. All models were adjusted for the same covariates as in the primary analysis. For ADL outcomes, the reported odds ratios correspond to the Class 3 \u0026times; Time interaction term, representing differences in the rate of change in ADL limitation risk over the follow-up period rather than baseline group differences.\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eIn analyses restricted to participants with higher classification certainty, the Class 3 interaction estimates in the depressive symptom models were broadly similar to those in the primary analysis (e.g., β\u0026thinsp;=\u0026thinsp;0.102), although confidence intervals widened in some specifications because of smaller sample size. In the ADL limitation models, point estimates for some classes increased as the maximum posterior probability (MaxPP) threshold was raised (e.g., OR\u0026thinsp;\u0026asymp;\u0026thinsp;1.26), but the confidence intervals continued to include the null.\u003c/p\u003e \u003cp\u003eOverall, the direction of the estimates was generally consistent across sensitivity analyses. Although statistical significance varied across some specifications, the results did not materially differ from the primary findings.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec17\" class=\"Section2\"\u003e \u003ch2\u003e3.6 Exploratory analysis of pain distribution and mediation\u003c/h2\u003e \u003cp\u003eBecause the musculoskeletal\u0026ndash;digestive class (Class 3) showed a steeper increase in depressive symptoms over follow-up, we further examined pain characteristics measured in 2013 and their potential relevance to later depressive symptoms.\u003c/p\u003e \u003cp\u003eAs shown in Table\u0026nbsp;\u003cspan refid=\"Tab8\" class=\"InternalRef\"\u003e8\u003c/span\u003e, overall pain prevalence, pain severity, and number of pain sites differed across classes. Class 3 had the highest overall pain prevalence (40.2%). It also showed a heavier overall pain burden in terms of pain severity and number of pain sites (both P\u0026thinsp;\u0026lt;\u0026thinsp;0.05). By contrast, no significant between-class differences were observed for single-site pain indicators, including stomach, waist, and knee pain. Although the overall distributions of pain severity and number of pain sites differed across classes, their medians and interquartile ranges were the same, suggesting that the differences were more apparent in the distribution as a whole than in central tendency.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab8\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 8\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003ePain characteristics in 2013 across latent multimorbidity classes (N\u0026thinsp;=\u0026thinsp;12,585)\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"6\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePain indicators\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eClass 4 (Relatively healthy)\u003c/p\u003e \u003cp\u003e(n\u0026thinsp;=\u0026thinsp;9,698)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eClass 1 (Cardiometabolic)\u003c/p\u003e \u003cp\u003e(n\u0026thinsp;=\u0026thinsp;1,303)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eClass 2 (Respiratory)\u003c/p\u003e \u003cp\u003e(n\u0026thinsp;=\u0026thinsp;302)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eClass 3 (Musculoskeletal\u0026ndash;digestive)\u003c/p\u003e \u003cp\u003e(n\u0026thinsp;=\u0026thinsp;1,282)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eP value\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eOverall pain prevalence (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e36.7\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e34.5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e34.8\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e40.2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.017\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSite-specific pain prevalence (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eStomach\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e4.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e3.9\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e3.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e5.3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.127\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLower back\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e15.9\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e14.7\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e14.2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e18.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.094\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eKnees\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e9.6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e8.5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e8.6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e10.5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.369\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNumber of pain sites, median [IQR]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0 [0,1]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0 [0,1]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0 [0,1]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0 [0,1]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.017\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePain severity (Median [IQR])\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0 [0,1]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0 [0,1]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0 [0,1]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0 [0,1]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"6\"\u003eNote: Categorical variables are presented as percentages, and ordinal or count variables as median [IQR]. Between-class differences were assessed using chi-square tests for categorical variables and Kruskal\u0026ndash;Wallis tests for pain severity and number of pain sites. The total sample (N\u0026thinsp;=\u0026thinsp;12,585) included participants in the longitudinal cohort who had at least one depressive outcome record in 2015 or 2018 and had available pain measures in 2013. Although the median and IQR were identical across classes, the Kruskal\u0026ndash;Wallis test indicated differences in the overall distributions.\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eWe then examined whether pain indicators measured in 2013 statistically mediated the association between baseline multimorbidity class and depressive symptom scores in 2015 and 2018. Across all three pain measures\u0026mdash;any pain, pain severity, and number of pain sites\u0026mdash;pain in 2013 was associated with higher subsequent depressive scores. However, the association between Class 3 and these pain indicators was modest, and no indirect effect reached statistical significance. For example, when number of pain sites was used as the mediator, the average causal mediation effect (ACME) was 0.032 (95% CI: \u0026minus;0.012 to 0.081; P\u0026thinsp;=\u0026thinsp;0.190).\u003c/p\u003e \u003cp\u003eOverall, these findings suggest that pain may reflect a higher symptom burden in the musculoskeletal\u0026ndash;digestive class and may be related to subsequent depressive symptoms. Under the current model specification, however, no clear statistical mediation effect was observed.\u003c/p\u003e \u003c/div\u003e"},{"header":"4. Discussion","content":"\u003cp\u003eMultimorbidity is increasingly recognized as a heterogeneous construct rather than a simple accumulation of chronic conditions[\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e, \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e]. Much of the existing literature has relied on disease counts or cross-sectional clustering, with less attention to how baseline multimorbidity patterns relate to subsequent psychological and functional outcomes over time[\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e]. In China, associations between multimorbidity patterns and depression or functional impairment have been reported, but longitudinal evidence remains limited[\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e]. Using a nationally representative cohort, we identified four clinically interpretable multimorbidity patterns and found that depressive symptom trajectories differed across classes, whereas the rate of change in ADL limitation did not. These findings support a structure-oriented perspective on multimorbidity and suggest that psychological and functional outcomes may not evolve in the same way across multimorbidity patterns[\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e, \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eThe main finding of this study was that the musculoskeletal\u0026ndash;digestive pattern (Class 3) was associated with a steeper increase in depressive symptoms over time, and this pattern was generally consistent across sensitivity analyses. Baseline depressive symptom levels did not differ significantly between Class 3 and the relatively healthy class, but their trajectories diverged during follow-up. This suggests that multimorbidity structure may be relevant not only to baseline burden, but also to subsequent change in mental health[\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e]. The finding is also consistent with previous studies showing that specific multimorbidity patterns are associated with later-life depression and poorer mental health outcomes[\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eOne possible explanation is symptom burden. Persistent pain, gastrointestinal discomfort, and related symptoms may affect mood through daily disruption, reduced activity, and lower social participation[\u003cspan additionalcitationids=\"CR30 CR31 CR32\" citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e]. By contrast, cardiometabolic conditions may be less symptomatically salient at earlier stages and may be managed through more standardized care pathways, which could partly attenuate their psychological impact[\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e, \u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e]. Previous studies have linked musculoskeletal disease and chronic pain to depression[\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e], and musculoskeletal-dominant multimorbidity patterns have also been associated with poorer quality of life and mental health[\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e, \u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e]. Our findings extend this literature by suggesting that the difference may lie in the trajectory of depressive symptoms over time rather than in baseline depressive levels alone[\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e, \u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e]. More broadly, they are consistent with the view that different combinations of chronic conditions may reflect different levels of physiological burden, symptom experience, and care needs, which may in turn shape subsequent psychological outcomes[\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e, \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e, \u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eThe exploratory mediation analysis showed that pain was associated with higher subsequent depressive symptom scores, but no clear statistical indirect effect was observed for the musculoskeletal\u0026ndash;digestive pattern under the current model specification. Pain therefore appears more consistent with a symptom-burden correlate than with a confirmed mediator in this analysis.\u003c/p\u003e \u003cp\u003eEven so, pain may still be relevant to the association between this pattern and later depressive symptoms. Within a biopsychosocial framework, pain may influence mood through disrupted sleep and a possible pain\u0026ndash;insomnia\u0026ndash;depression cycle[\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e, \u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e], reduced activity and social participation accompanied by greater isolation or perceived loss[\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e], and increased bodily vigilance related to gastrointestinal symptoms[\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e]. These possibilities are broadly consistent with the higher overall pain burden observed in the musculoskeletal\u0026ndash;digestive class.\u003c/p\u003e \u003cp\u003eOther mechanisms, including systemic inflammation and gut\u0026ndash;brain pathways, have also been proposed in the literature, but they were not measured in the present study and therefore remain speculative[\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e]. Further studies incorporating inflammatory markers, sleep measures, medication use, and repeated pain assessments may help clarify the behavioural and biological pathways involved.\u003c/p\u003e \u003cp\u003eUnlike the findings for depressive symptoms, we did not observe significant between-class differences in the rate of change in ADL limitation, although some point estimates were higher in the musculoskeletal\u0026ndash;digestive and cardiometabolic classes. Several explanations are possible. First, change in basic ADL may emerge more slowly or nonlinearly than change in depressive symptoms, and the follow-up period may not have been long enough to detect differences in slopes[\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e, \u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e]. Earlier or subtler functional changes may also be better captured by more sensitive measures such as instrumental activities of daily living (IADL)[\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e]. Second, attrition due to loss to follow-up or mortality may have attenuated associations, particularly if participants with poorer health were less likely to remain under observation. Classification uncertainty may also have reduced the contrast between classes, although point estimates tended to increase in some higher-certainty analyses[\u003cspan additionalcitationids=\"CR39\" citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e]. Third, in community-dwelling populations, compensatory strategies may allow individuals to maintain basic ADL despite declines in intrinsic capacity[\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eTaken together, these findings suggest that multimorbidity structure may be more informative for depressive symptom trajectories than for short- to mid-term change in basic ADL. Within the present follow-up period, the increase in ADL limitation appeared to reflect a more general age-related pattern across classes.\u003c/p\u003e \u003cp\u003eWe also observed opposite directional associations of BMI with depressive symptom change and ADL limitation risk. This pattern is broadly consistent with discussions of an \u0026ldquo;obesity paradox\u0026rdquo; in older adults, whereby higher BMI may reflect greater nutritional reserve and lower wasting risk, while also increasing the risk of functional limitation through mechanical and cardiopulmonary burden[\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e, \u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e, \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e]. At the same time, BMI is an imperfect marker in older populations because it does not distinguish fat mass from lean mass and may be influenced by illness-related weight loss or reverse causation. This finding should therefore be interpreted cautiously. Future studies incorporating body composition, sarcopenia, and physical performance measures may provide a clearer picture.\u003c/p\u003e \u003cp\u003eFrom a public health and primary care perspective, these findings suggest that multimorbidity management may benefit from moving beyond simple disease counts. Stratification based only on the number of conditions may overlook clinically relevant heterogeneity. In particular, patterns characterized by higher symptom burden, such as the musculoskeletal\u0026ndash;digestive class, may warrant closer attention to mental health risk and symptom management[\u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e, \u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e44\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eSeveral practical implications follow. First, mental health screening could be more explicitly incorporated into routine chronic disease management, particularly for individuals with chronic musculoskeletal pain and gastrointestinal conditions[\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e]. Second, care models that integrate symptom control and psychological support may be useful, especially where pain management, primary care, and mental health services are delivered separately[\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e]. Third, the absence of clear between-class differences in ADL slope, alongside the earlier divergence in depressive symptoms, may indicate a window for earlier intervention before functional decline becomes more apparent. In resource-limited settings, simple rule-based identification strategies and longitudinal follow-up within primary care or electronic health record systems may help target support more efficiently[\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e, \u003cspan additionalcitationids=\"CR44\" citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e45\u003c/span\u003e].\u003c/p\u003e \u003cp\u003e \u003cb\u003eLimitations\u003c/b\u003e \u003c/p\u003e \u003cp\u003eSeveral limitations should be considered when interpreting these findings. First, classification quality in the latent class model was moderate (entropy\u0026thinsp;=\u0026thinsp;0.702), indicating some uncertainty in individual class assignment. Although sensitivity analyses using MaxPP thresholds showed generally similar results, the latent classes are better interpreted as population-level multimorbidity structures than as fixed labels at the individual level. Second, ADL limitation was self-reported and may have been less sensitive to early or subtle functional change. In addition, the respiratory-dominant class was relatively small, which reduced the precision of estimates for this pattern. Third, attrition due to loss to follow-up or mortality may have introduced selection bias if it was related to health status. Although missing data were handled using multiple imputation and maximum likelihood under a missing at random (MAR) assumption, missing not at random (MNAR) mechanisms cannot be ruled out. Finally, the exploratory mechanistic analyses were based on an observational design and remained subject to unmeasured confounding and other assumptions required for mediation analysis. These results should therefore be interpreted as exploratory evidence on possible pathways rather than confirmation of a causal mechanism. Future studies with longer follow-up, repeated symptom assessments, and more detailed behavioural and biological measures may help clarify these pathways.\u003c/p\u003e"},{"header":"Conclusion","content":"\u003cp\u003eUsing nationally representative longitudinal data, this study identified four clinically interpretable multimorbidity patterns and showed that different multimorbidity structures were associated with different longitudinal patterns of depressive symptoms and ADL limitation. The musculoskeletal\u0026ndash;digestive pattern was the only class associated with a faster increase in depressive symptoms over time. By contrast, although the risk of ADL limitation increased across all classes, no clear between-class differences in the rate of change were observed under the current analytical framework.\u003c/p\u003e \u003cp\u003eExploratory analyses further suggested that pain may reflect a higher symptom burden in the musculoskeletal\u0026ndash;digestive pattern and may be related to subsequent depressive symptoms, although no clear mediation effect was observed. Overall, these findings support the value of considering multimorbidity structure alongside disease counts in risk assessment. In public health and primary care settings, identifying patterns with higher symptom burden may help inform mental health screening and follow-up among middle-aged and older adults.\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003cdiv class=\"DefinitionList\"\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eADL\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eActivities of daily living\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eBMI\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eBody mass index\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eBLRT\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eBootstrap likelihood ratio test\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eCES-D-10\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003e10-item Center for Epidemiologic Studies Depression Scale\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eCHARLS\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eChina Health and Retirement Longitudinal Study\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eCI\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eConfidence interval\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eFIML\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eFull information maximum likelihood\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eGLMM\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eGeneralized linear mixed-effects model\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eIADL\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eInstrumental activities of daily living\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eLCA\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eLatent class analysis\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eLMM\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eLinear mixed-effects model\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eLMR\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eLo\u0026ndash;Mendell\u0026ndash;Rubin adjusted likelihood ratio test\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eMAR\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eMissing at random\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eMaxPP\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eMaximum posterior probability\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eMICE\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eMultiple imputation by chained equations\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eMNAR\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eMissing not at random\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eOR\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eOdds ratio\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003ePHQ\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003ePatient Health Questionnaire\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eSD\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eStandard deviation\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003c/div\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eEthics approval and consent to participate\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe CHARLS study was approved by the Biomedical Ethics Review Committee of Peking University (IRB00001052-11015). The study was conducted in accordance with relevant institutional and national regulations. All participants provided written informed consent prior to participation. The study was conducted in accordance with the Declaration of Helsinki.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\u003ch2\u003eClinical trial number\u003c/h2\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\u003ch2\u003eCompeting interests\u003c/h2\u003e\n\u003cp\u003eThe authors declare that they have no competing interests.\u003c/p\u003e\n\u003ch2\u003eFunding\u003c/h2\u003e\n\u003cp\u003eThis study was supported by the Xinjiang Uygur Autonomous Region Social Science Fund Project 21BRK116. This study has also been approved as a general project of the Xinjiang Nursing Society in 2025, with project number 2025XH38.\u003c/p\u003e\n\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\n\u003cp\u003eKaige Gao, Yanjie Zhao, Xuehong Zhang, Kanbao Su, Jia Jia and Yuezhen Xu wrote the main manuscript text and prepared figures and tables. All authors reviewed the manuscript.\u003c/p\u003e\n\u003ch2\u003eAcknowledgement\u003c/h2\u003e\n\u003cp\u003eThank Charls researchers and investigators for selflessly providing publicly available data\u003c/p\u003e\n\u003ch2\u003eData Availability\u003c/h2\u003e\n\u003cp\u003eThe datasets analyzed in this study are available in the China Health and Retirement Longitudinal Study (CHARLS) repository. Access to the data requires online registration and approval through the official CHARLS website.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eGulliford MC, Green JM. Is multimorbidity a useful concept for public health? Lancet Public Health. 2024;9(4):e210\u0026ndash;1.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAl-Chalabi S, Sinha S, Kalra PA. Enhancing clinical service design for multimorbidity management: A comprehensive approach to joined-up care for diabetes, chronic kidney disease, and heart failure. Diabet Med. 2025;42(2):e15403.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNguyen H, Moreno-Agostino D, Chua KC, Vitoratou S, Prina AM. Trajectories of healthy ageing among older adults with multimorbidity: A growth mixture model using harmonised data from eight ATHLOS cohorts. PLoS ONE. 2021;16(4):e0248844.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCezard G, McHale CT, Sullivan F, Bowles JKF, Keenan K. Studying trajectories of multimorbidity: a systematic scoping review of longitudinal approaches and evidence. BMJ Open. 2021;11(11):e048485.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhang Y, Zhou Y, Kivim\u0026auml;ki M, Holt-Lunstad J, Carrillo-Larco RM, Xu X. Cardiometabolic multimorbidity, social activity, and joint trajectories of physical disability, depressive symptom, and cognitive function in mid-to-late life: a multicohort study. BMC Med. 2025;23(1):577.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDong X, Ma Y, Zhang H, Wang P. Individual-level transitions between chronic disease multimorbidity clusters and the risk of five-year mortality in longitudinal cohort of Chinese middle-aged and older adults. Aging Clin Exp Res. 2025;37(1):216.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBayes-Marin I, Sanchez-Niubo A, Egea-Cort\u0026eacute;s L, Nguyen H, Prina M, Fern\u0026aacute;ndez D, Haro JM, Olaya B. Multimorbidity patterns in low-middle and high income regions: a multiregion latent class analysis using ATHLOS harmonised cohorts. BMJ Open. 2020;10(7):e034441.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSinha P, Calfee CS, Delucchi KL. Practitioner's Guide to Latent Class Analysis: Methodological Considerations and Common Pitfalls. Crit Care Med. 2021;49(1):e63\u0026ndash;79.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLiu Q, Lu J. Prevalence of multimorbidity and its relationship with socioeconomic status among Chinese older adults over time. PLoS ONE. 2025;20(4):e0322042.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHu X, Li S, Wei Z, Wu D, Meng L, Li J, Xu J, Zhang L, Ma Q, Li H, et al. Prevalence and pattern of multimorbidity in China: a cross-sectional study of 224,142 adults over 60 years old. Front Public Health. 2024;12:1349418.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLong C, Huang J, Liu D, Liu C, Wu M, Wu H, Deng J, Zhang Y, Shi L, Cui Y. Prevalence, combination patterns, and quality of life factors of multimorbidity among older adults in southern China based on the health ecological model. J Glob Health. 2025;15:04215.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhou L, Wang W, Ma X. The bidirectional association between the disability in activities of daily living and depression: a longitudinal study in Chinese middle-aged and older adults. BMC Public Health. 2024;24(1):1884.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhu X, Wang Y, Luo Y, Ding R, Shi Z, He P. Bidirectional, longitudinal associations between depressive symptoms and IADL/ADL disability in older adults in China: a national cohort study. BMC Geriatr. 2024;24(1):659.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhao Y, Hu Y, Smith JP, Strauss J, Yang G. Cohort profile: the China Health and Retirement Longitudinal Study (CHARLS). Int J Epidemiol. 2014;43(1):61\u0026ndash;8.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNing M, Zhang Q, Yang M. Comparison of self-reported and biomedical data on hypertension and diabetes: findings from the China Health and Retirement Longitudinal Study (CHARLS). BMJ Open. 2016;6(1):e009836.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eXie D, Wang J. Comparison of self-reports and biomedical measurements on hypertension and diabetes among older adults in China. BMC Public Health. 2020;20(1):1664.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhao Y, Strauss J, Yang G, Giles J, Hu PP, Hu Y, Lei X, Liu M, Park A, Smith JP, et al. China Health and Retirement Longitudinal Study: 2011\u0026ndash;2012 National Baseline User\u0026rsquo;s Guide. Beijing, China: National School of Development, Peking University; 2013.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003evan Buuren S. Flexible Imputation of Missing Data. 2nd ed. New York: Chapman and Hall/CRC; 2018.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRubin DB. Multiple Imputation for Nonresponse in Surveys. New York: Wiley; 1987.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eImai K, Keele L, Tingley D. A general approach to causal mediation analysis. Psychol Methods. 2010;15(4):309\u0026ndash;34.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eImai K, Keele L, Yamamoto T. Identification, Inference and Sensitivity Analysis for Causal Mediation Effects. Stat Sci. 2010;25(1):51\u0026ndash;71.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOgaz-Gonz\u0026aacute;lez R, Zou Q, Du Y, Guti\u0026eacute;rrez-Robledo LM, Escamilla-Santiago R, L\u0026oacute;pez-Cervantes M, Corpeleijn E. Multimorbidity patterns in older adults from Northern Netherlands: comparing factor analysis and latent class analysis solutions. BMC Geriatr. 2025;25(1):381.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLiu Q, Lin S, Yin L, Liu W, Zhu X, Li Z, Shen Y, Feng B. Identification and trajectory of multimorbidity patterns among older people in China: a longitudinal study based on the China health and retirement longitudinal study 2011\u0026ndash;2020 data. Front Public Health. 2025;13:1597224.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBao L, Zhang W, Zhang Y, Wang M, Chen Y, Shan X, Xing Y. The relationships between multimorbidity, depressive symptoms, health service utilization, and activities of daily living among the elderly in China. PLoS ONE. 2025;20(10):e0333923.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWu Y, Su B, Chen C, Zhao Y, Zhong P, Zheng X. Urban-rural disparities in the prevalence and trends of depressive symptoms among Chinese elderly and their associated factors. J Affect Disord. 2023;340:258\u0026ndash;68.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZheng DD, Loewenstein DA, Christ SL, Feaster DJ, Lam BL, McCollister KE, Curiel-Cid RE, Lee DJ. Multimorbidity patterns and their relationship to mortality in the US older adult population. PLoS ONE. 2021;16(1):e0245053.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRonaldson A, Arias de la Torre J, Bendayan R, Yadegarfar ME, Rhead R, Douiri A, Armstrong D, Hatch S, Hotopf M, Dregan A. Physical multimorbidity, depressive symptoms, and social participation in adults over 50 years of age: findings from the English Longitudinal Study of Ageing. Aging Ment Health. 2023;27(1):43\u0026ndash;53.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWang X, Zheng N, Yin M. Multimorbidity Patterns and Depression: Bridging Epidemiological Associations with Predictive Analytics for Risk Stratification. \u003cem\u003eHealthcare (Basel)\u003c/em\u003e 2025, 13(12).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLehti TE, \u0026Ouml;hman H, Knuutila M, Kautiainen H, Karppinen H, Tilvis R, Strandberg T, Pitk\u0026auml;l\u0026auml; KH. Symptom burden in community-dwelling older people: temporal trends in the Helsinki Aging Study. Aging Clin Exp Res. 2021;33(11):3065\u0026ndash;71.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMcKay MA, Bernhardt P, O'Connor M, Leveille S. Characterizing symptom burden among community-dwelling older adults. Geriatr Nurs. 2026;69:103899.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHuang MH, Wang YP, Wu PS, Chan YE, Cheng CM, Yang CH, Tsai SJ, Lu CL, Tsai CF. Association between gastrointestinal symptoms and depression among older adults in Taiwan: A cross-sectional study. J Chin Med Assoc. 2021;84(3):331\u0026ndash;5.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSharpe L, McDonald S, Correia H, Raue PJ, Meade T, Nicholas M, Arean P. Pain severity predicts depressive symptoms over and above individual illnesses and multimorbidity in older adults. BMC Psychiatry. 2017;17(1):166.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYe X, Wang X. Associations of multimorbidity with body pain, sleep duration, and depression among middle-aged and older adults in China. Health Qual Life Outcomes. 2024;22(1):23.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKroenke K, Wu J, Bair MJ, Krebs EE, Damush TM, Tu W. Reciprocal relationship between pain and depression: a 12-month longitudinal analysis in primary care. J Pain. 2011;12(9):964\u0026ndash;73.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHo HE, Yeh CJ, Cheng-Chung Wei J, Chu WM, Lee MC. Association between multimorbidity patterns and incident depression among older adults in Taiwan: the role of social participation. BMC Geriatr. 2023;23(1):177.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMcClain AC, Xiao RS, Tucker KL, Falc\u0026oacute;n LM, Mattei J. Depressive symptoms and allostatic load have a bidirectional association among Puerto Rican older adults. Psychol Med. 2022;52(14):3073\u0026ndash;85.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGriffin SC, Ravyts SG, Bourchtein E, Ulmer CS, Leggett MK, Dzierzewski JM, Calhoun PS. Sleep disturbance and pain in U.S. adults over 50: evidence for reciprocal, longitudinal effects. Sleep Med. 2021;86:32\u0026ndash;9.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLingying W, Hong Z, Hongxiu C, Ziyi H, Mei F, Menglin T, Xiuying H. Association of body mass index with disability in activities of daily living in older adults: a systematic review of the literature based on longitudinal data. BMC Public Health. 2025;25(1):6.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLi X, Feng H, Chen Q. Social participation patterns and associations with subsequent cognitive function in older adults with cognitive impairment: a latent class analysis. Front Med (Lausanne). 2025;12:1493359.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYin F, Fan X, Chen J, Wu X. Leisure activity patterns and health vulnerability among older adults in China: a nationwide analysis of urban-rural differences based on CHARLS data. BMC Geriatr 2026, 26(1).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJia H, Lubetkin EI. Association between self-reported body mass index and active life expectancy in a large community-dwelling sample of older U.S. adults. BMC Geriatr. 2022;22(1):310.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePan H, Hao Y. Body Mass Index and multidimensional health in Chinese older adults: a moderated mediation analysis of urban-rural residence, physical activity, and depression. Front Public Health. 2025;13:1650975.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMarengoni A, Tazzeo C, Calder\u0026oacute;n-Larra\u0026ntilde;aga A, Roso-Llorach A, Onder G, Zucchelli A, Rizzuto D, Vetrano DL. Multimorbidity Patterns and 6-Year Risk of Institutionalization in Older Persons: The Role of Social Formal and Informal Care. J Am Med Dir Assoc. 2021;22(10):2184\u0026ndash;e21892181.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBarrio-Cortes J, Casta\u0026ntilde;o-Reguillo A, Benito-S\u0026aacute;nchez B, Beca-Mart\u0026iacute;nez MT, Ruiz-Zaldibar C. Utilization of Primary Healthcare Services in Patients with Multimorbidity According to Their Risk Level by Adjusted Morbidity Groups: A Cross-Sectional Study in Chamart\u0026iacute;n District (Madrid). Healthc (Basel) 2024, 12(2).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSmith SM, Wallace E, Clyne B, Boland F, Fortin M. Interventions for improving outcomes in patients with multimorbidity in primary care and community setting: a systematic review. Syst Rev. 2021;10(1):271.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"bmc-public-health","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"pubh","sideBox":"Learn more about [BMC Public Health](http://bmcpublichealth.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/pubh/default.aspx","title":"BMC Public Health","twitterHandle":"@BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"Middle-aged and older adults, Multimorbidity, Latent class analysis, Depressive symptoms, Activities of daily living, China","lastPublishedDoi":"10.21203/rs.3.rs-9085989/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9085989/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cb\u003eBackground\u003c/b\u003e\u003c/p\u003e \u003cp\u003eMultimorbidity is heterogeneous, but longitudinal evidence from China on how specific disease patterns relate to changes in depressive symptoms and functional limitations remains limited. This study aimed to identify baseline multimorbidity patterns and examine their longitudinal associations with trajectories of depressive symptoms and activities of daily living (ADL) limitations in a nationally representative cohort.\u003c/p\u003e\u003cp\u003e\u003cb\u003eMethods\u003c/b\u003e\u003c/p\u003e \u003cp\u003eData were drawn from the China Health and Retirement Longitudinal Study (CHARLS), 2011\u0026ndash;2018. Latent class analysis (LCA) was used to identify baseline (2011) multimorbidity patterns based on 12 chronic conditions. Linear and generalized linear mixed-effects models were applied to assess the associations between baseline patterns and changes in depressive symptoms and the risk of ADL limitation from 2013 to 2018. Exploratory mediation analysis examined whether pain statistically mediated the association between multimorbidity patterns and subsequent depressive symptoms.\u003c/p\u003e\u003cp\u003e\u003cb\u003eResults\u003c/b\u003e\u003c/p\u003e \u003cp\u003eFour multimorbidity patterns were identified: cardiometabolic (Class 1), respiratory-dominant (Class 2), musculoskeletal\u0026ndash;digestive (Class 3), and relatively healthy (Class 4). Compared with Class 4, Class 3 showed a steeper increase in depressive symptom scores over time (β\u0026thinsp;=\u0026thinsp;0.099, P\u0026thinsp;=\u0026thinsp;0.020). Findings were directionally consistent in sensitivity analyses, including the random-slope model (β\u0026thinsp;=\u0026thinsp;0.096) and the high-classification-certainty subsample (β\u0026thinsp;=\u0026thinsp;0.102). In exploratory mediation analyses, pain indicators measured in 2013 were associated with higher subsequent depressive scores, but no clear statistically indirect effect was observed for Class 3. The risk of ADL limitation increased over time (P\u0026thinsp;\u0026lt;\u0026thinsp;0.001), but no significant differences in the rate of change were found across patterns (all class-by-time interaction terms P\u0026thinsp;\u0026gt;\u0026thinsp;0.05).\u003c/p\u003e\u003cp\u003e\u003cb\u003eConclusions\u003c/b\u003e\u003c/p\u003e \u003cp\u003eDistinct multimorbidity patterns were differentially associated with depressive symptom trajectories among middle-aged and older adults in China. The musculoskeletal\u0026ndash;digestive pattern was associated with a faster increase in depressive symptoms over time. Pain may reflect an important symptom-burden correlate of this pattern, but no clear mediation evidence was observed under the current model specification. These findings suggest that mental health screening and symptom-oriented assessment may be particularly relevant for high-symptom-burden multimorbidity patterns in primary care.\u003c/p\u003e","manuscriptTitle":"Latent multimorbidity patterns and their longitudinal associations with depressive symptom trajectories and ADL limitations among middle-aged and older adults in China: a longitudinal analysis of the CHARLS","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-04-02 15:16:46","doi":"10.21203/rs.3.rs-9085989/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2026-05-04T10:59:05+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-04-24T22:18:59+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-04-17T14:47:26+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"292625795412663983823775038285451598014","date":"2026-04-03T18:25:51+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"160109025925913750182195561863731706461","date":"2026-03-30T21:41:29+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"308547681389153176807674541361938630331","date":"2026-03-29T16:02:16+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2026-03-29T14:34:07+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2026-03-16T09:59:29+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2026-03-13T11:30:50+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2026-03-13T11:30:13+00:00","index":"","fulltext":""},{"type":"submitted","content":"BMC Public Health","date":"2026-03-10T15:47:18+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"bmc-public-health","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"pubh","sideBox":"Learn more about [BMC Public Health](http://bmcpublichealth.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/pubh/default.aspx","title":"BMC Public Health","twitterHandle":"@BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"933c0bd9-3592-422c-9644-807c351d42c6","owner":[],"postedDate":"April 2nd, 2026","published":true,"recentEditorialEvents":[{"type":"decision","content":"Revision requested","date":"2026-05-04T10:59:05+00:00","index":"","fulltext":""}],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2026-05-17T11:53:22+00:00","versionOfRecord":[],"versionCreatedAt":"2026-04-02 15:16:46","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-9085989","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9085989","identity":"rs-9085989","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}