Abstract
Background
Population aging and declining fertility have transformed family life courses, yet their long-term implications for women’s health remain insufficiently understood. This study examined whether distinct marital and fertility trajectories were associated with subsequent changes in depressive symptoms and physical functional limitations among Chinese women.
Methods
We used data from the China Health and Retirement Longitudinal Study (CHARLS), including 5,871 women aged 45–65 years at baseline. Retrospective marital and fertility histories from ages 15 to 49 were reconstructed using sequence-based methods and grouped into trajectory typologies through Ward’s clustering. Depressive symptoms were assessed using the CES-D-10, and physical functional limitations were measured using the Nagi Disability Scale. Associations with longitudinal health outcomes were estimated using linear mixed-effects models. Robustness analyses additionally adjusted for childhood contexts, and sensitivity analyses restricted the sample to women with at least two valid outcome observations.
Results
Four trajectory groups were identified: Stable Marriage & Two-Child (40.0%), Early Marriage & High-Parity (43.3%), Stable Marriage & One-Child (15.6%), and Predominantly Married & Childless (1.2%). Compared with the Stable Marriage & Two-Child group, the Early Marriage & High-Parity group had higher baseline depressive symptoms (β = 0.858, 95% CI: 0.560 to 1.157) and higher physical functional limitations (β = 0.206, 95% CI: 0.143 to 0.269), with a faster increase in physical functional limitations over time (β = 0.025, 95% CI: 0.012 to 0.038). The Stable Marriage & One-Child group had lower baseline depressive symptoms (β = -0.858, 95% CI: -1.271 to -0.445) and lower physical functional limitations (β = -0.163, 95% CI: -0.251 to -0.076), with a marginally slower increase in physical functional limitations (β = -0.018, 95% CI: -0.036 to 0.000). The Predominantly Married & Childless group did not differ significantly from the reference group. These associations were broadly consistent in robustness and sensitivity analyses.
Conclusions
Women’s marital and fertility trajectories were associated with later depressive symptoms and physical functional limitations from midlife to early old age. Early marriage and high parity were linked to poorer depressive and functional outcomes, whereas the Stable Marriage & One-Child trajectory was associated with relatively better depressive and functional outcomes. These findings suggest that the timing and structure of family formation may have lasting implications for women’s later-life depressive symptoms and physical functional limitations.
Supplementary Information
The online version contains supplementary material available at https://doi.org/10.1186/s12889-026-28482-y.
Keywords: Life course, Family trajectories, Sequence analysis, Depression, Physical functional limitations, CHARLS
Introduction
Family life-course and health in midlife to early old age
As life expectancy rises and fertility rates continue to fall, family structures in many societies are becoming smaller and more vertically extended, with fewer children and a greater share of kinship ties concentrated across generations rather than within large sibling networks [1]. For women, the cumulative consequences of marital and reproductive histories may shape health well beyond the years in which these events occur [2]. Understanding how family life-course patterns are linked to midlife-to-early-old-age health trajectories is an important public health question, particularly in settings where reproductive histories and kin-based support remain central to women’s later-life resources [3].
Maternal depletion and social support
The long-term health implications of parity and marital history have generated substantial debate within life-course epidemiology [4]. One line of argument is the maternal depletion perspective, which suggests that repeated pregnancies, short birth intervals, and prolonged childrearing demands may impose physiological strain over time and contribute to poorer later health [5, 6]. These effects may begin early in adulthood, as repeated pregnancies, early marriage, and the demands of childrearing may cumulatively constrain women’s health, education, and occupational opportunities [7, 8]. Women who marry early and bear more children may therefore experience a longer accumulation of reproductive and caregiving demands, which could contribute to more disadvantaged health trajectories over time [9]. A related but distinct line of reasoning emphasizes social support: children and marital partners may provide economic resources, practical assistance, and emotional support that help buffer stress and maintain health in later adulthood [10]. Marital stability may also matter earlier in the life course by providing economic security and reducing chronic stress during the reproductive years, not only in later adulthood when care needs become more salient [11]. Conversely, women with stable marriages and fewer children may have had greater access to these resources throughout adulthood, contributing to more favorable health trajectories over time [3]. From this perspective, family size and family composition may matter not only because of reproductive burden, but also because they shape the availability of kin-based support over the life course [10].
The Chinese context and research gaps
China provides a particularly informative setting in which to examine these issues [12]. Family relations have historically been shaped by strong norms of filial obligation, with adult children often serving as the main source of old-age support [13, 14]. At the same time, fertility patterns have been closely tied to historical and socioeconomic stratification, with higher-parity families disproportionately concentrated in rural and economically disadvantaged contexts [15, 16]. These conditions matter because they create the possibility that family trajectories are entangled with broader social and occupational pathways, making confounding an important concern [17, 18]. In addition, although formal social protection has expanded in recent decades, family-based support remains central for many older women, especially in settings where long-term care resources are uneven [19, 20]. Together, these features make China a useful context for examining how marital and fertility trajectories relate to later health [21]. A further limitation of prior work is that family circumstances are often captured using single-point indicators such as current marital status or completed parity [22]. These measures do not reflect the timing, sequencing, and duration of family experiences across adulthood [23]. More recently, sequence-based approaches have made it possible to study family life-course patterns more directly, but evidence from China remains limited, particularly for the joint long-term implications of marital and fertility trajectories for depressive symptoms and physical functional limitations [24–26].
The present study
To address this gap, we used data from the China Health and Retirement Longitudinal Study (CHARLS) and sequence-based methods to reconstruct marital and fertility histories from ages 15 to 49 among women aged 45 to 65 at baseline [22]. We then examined whether distinct family life-course trajectories were associated with subsequent changes in depressive symptoms and physical functional limitations over a 7-year follow-up [27, 28]. We also assessed whether associations remained robust after adjustment for childhood contexts and in sensitivity analyses restricted to participants with at least two valid outcome observations [29, 30]. We did not assume a single dominant mechanism; instead, we allowed for the possibility that each trajectory type might carry a distinct combination of reproductive burden, social support, and socioeconomic disadvantage.
Methods
Study design and participants
We used data from the CHARLS, a nationally representative longitudinal survey of Chinese adults aged 45 and older [31]. To capture the full family life-course, we retrospectively extracted detailed marital and fertility histories from the 2014 CHARLS Life Course Survey. We then linked these historical records to four longitudinal follow-up waves spanning seven years (2011, 2013, 2015, and 2018).
The analytical sample was restricted to women aged 45–65 at the 2011 baseline. The participant selection process is illustrated in Fig. 1. We excluded individuals who did not participate in the 2014 Life Course Survey, those with missing life-course data, and those whose first marriage or first birth occurred before age 18 (per the 1950 Marriage Law). Individuals with missing baseline analytical covariates were additionally excluded. This process yielded a final analytical sample of 5,871 women. For each health outcome, we included participants with at least one valid measurement across the four follow-up waves, yielding analytical samples of 5,827 women for depressive symptoms and 5,870 women for physical functional limitations [30].
Fig. 1.
Flow diagram of sample selection and analytic inclusion. The flow diagram shows the derivation of the final analytic sample from the original CHARLS cohort. Analyses were restricted to women aged 45–65 years at baseline who participated in the 2014 Life Course Survey and provided valid life-history information. Outcome-specific analytic samples were defined for depressive symptoms and physical functional limitations and included participants with at least one valid measurement across the four follow-up waves. Baseline characteristics of included and excluded participants are shown in Table S6
Exposure: multichannel sequence analysis (MCSA)
Our primary exposure was the family life-course trajectory from age 15 to 49. Although women whose first marriage or first birth occurred before age 18 were excluded, sequences were initiated at age 15 to retain the pre-marital and pre-childbearing period before age 18. We used sequence analysis to summarize the timing, order, and duration of marital and fertility experiences across this 35-year span, treating each woman’s life-course record as two synchronized state sequences [22].
For each participant, we constructed two parallel sequence channels: one for annual marital status and one for cumulative live births. Marital status was coded as unmarried, married, divorced, or widowed. Fertility history was coded as 0, 1, 2, 3, or ≥ 4 live births. The two channels were aligned annually from age 15 to 49, so that each person contributed one marital state and one fertility state at each age.
We calculated a multichannel dissimilarity matrix using the seqdistmc function in the TraMineR package. Distances were based on multichannel optimal matching (MC-OM), with substitution costs derived empirically from observed state transition rates and insertion-deletion costs set to 1.0 for both channels [22]. Equal weights were assigned to the two channels when computing the combined dissimilarity matrix.
Ward’s D2 hierarchical clustering was then applied to the multichannel dissimilarity matrix to identify family life-course trajectory groups [32]. To determine the optimal number of clusters, we evaluated solutions ranging from k = 2 to k = 8 using six complementary clustering quality metrics: Average Silhouette Width (ASW), Hubert’s Gamma (HG), Calinski-Harabasz Index (CH), Point Biserial Correlation (PBC), R², and Hubert’s C (HC). Cluster selection followed a majority-rule principle, with primary emphasis on ASW and substantive interpretability. The final trajectory clusters were used as the main exposure categories in all subsequent analyses. Cluster labels were assigned post hoc based on the dominant patterns in the marital and fertility state distributions and were intended to be descriptive rather than to imply formal thresholds for marriage timing or causal mechanisms.
Longitudinal outcomes
Two primary health outcomes were tracked over the 7-year follow-up. Depressive symptoms were measured using the 10-item Center for Epidemiologic Studies Depression Scale (CES-D-10), which assesses the frequency of symptoms experienced during the past week, including feeling depressed, fearful, lonely, and having restless sleep. Scores range from 0 to 30, with higher scores indicating more severe depressive symptoms [27]. Physical functional limitations were assessed using the six-item Nagi Disability Scale, which captures limitations in basic physical tasks and has demonstrated sensitivity to early functional decline in middle-aged populations [28]. The summary score (range 0–6) was based on six tasks: stooping, kneeling, or crouching; lifting or carrying approximately 5 kg; reaching arms above the shoulders; picking up a small coin from a table; walking five blocks; and climbing three flights of stairs. A higher score indicates more self-reported physical functional limitations. We interpreted the Nagi scale as a measure of functional limitations rather than as a direct measure of intrinsic biological physical health, because reported difficulty in physical tasks may also reflect environmental demands and available support.
Covariates
Baseline covariates included education (below upper secondary vs. upper secondary or above), residence (urban vs. rural), occupation (agricultural vs. non-agricultural), and birth cohort (1946–50, 1951–55, 1956–60, and 1961–66). These variables capture the main axes of socioeconomic stratification in China and account for historical and policy-related differences in family formation and fertility timing across generations. Chronological age was not included as a covariate given its collinearity with birth cohort. Later-life chronic conditions were not adjusted for in the primary models, as they may lie on the causal pathway linking family life-course trajectories to later-life health outcomes and their inclusion could introduce over-adjustment bias.
To explore potential pathways, we also examined several intergenerational and social factors measured at baseline, including: (1) intergenerational financial support (receiving financial transfers from offspring in the past year); (2) social participation (engagement in social or community activities in the past month); and (3) fitness activities (regular engagement in dancing or other physical exercises). Furthermore, childhood adversity before age 15 was assessed using three retrospective indicators: famine exposure [33], self-reported childhood health status (fair/poor vs. good), and family financial standing (poorer vs. same/better compared to others) [29, 34].
Statistical analysis and sensitivity tests
Linear mixed-effects models (LMM) with random intercepts and random slopes at the individual level were used to estimate longitudinal health trajectories over the 7-year follow-up [35]. Time was centered at the midpoint of follow-up. A quadratic time term was used for depressive symptoms to allow for nonlinear change over time, whereas a linear specification was used for functional limitations because change in this outcome was approximately linear over follow-up. The Stable Marriage & Two-Child group served as the reference category in all models. Trajectory × Time interaction terms were included to test whether the rate of change differed across family life-course groups. For depressive symptoms, a quadratic time term was additionally included as a main effect to allow for nonlinear change over time. We retained the original Nagi score in the primary models to preserve interpretability of the 0–6 functional limitation scale.
We conducted several robustness analyses: (1) adjusting for the childhood adversity variables (Table S2); (2) re-running models after excluding participants with baseline outcome scores greater than zero (CES-D-10 > 0 for depressive symptoms or Nagi > 0 for physical functional limitations) [36]; and (3) restricting the sample to women aged 45–60 at baseline to exclude potential outliers. Because the Nagi score was bounded between 0 and 6 and concentrated at the lower end of the scale, the second sensitivity analysis for physical functional limitations was restricted to women with Nagi scores of 0 at baseline. We then refitted the same linear mixed-effects model using all available repeated Nagi measurements from these women. This analysis assessed whether trajectory-by-time differences in functional limitations remained evident among participants who began follow-up with no reported functional limitations. To explore potential pathways, we conducted an exploratory analysis of the intergenerational and social factors across trajectory groups (Table S1). Categorical variables were compared using chi-square tests, continuous variables using one-way ANOVA, and logistic regression was used to estimate odds ratios (ORs) for binary indicators.
All analyses were performed using Stata version 19.5 and R version 4.5.2. A two-sided p-value < 0.05 was considered statistically significant.
Results
Baseline Characteristics and Family Trajectories
Of the final analytical cohort, 5,827 and 5,870 women contributed longitudinal data on depressive symptoms and physical functional limitations, respectively. Baseline characteristics by trajectory group are presented in Table 1. The mean baseline age was 54.18 years (SD = 5.91), with participants spanning four birth cohorts from 1946 to 1966. The cohort was characterized by relatively low socioeconomic status: 90.7% had education below upper secondary, 79.8% resided in rural areas, and 34.1% were engaged in agricultural occupations. Mean baseline depressive symptoms (CES-D-10) and physical functional limitations (Nagi scale) scores were 9.01 (SD = 6.50) and 1.04 (SD = 1.29), respectively. Baseline Nagi scores were concentrated at the lower end of the scale, with 2,762 of 5,870 women (47.1%) reporting no physical functional limitations at baseline.
Table 1.
Baseline characteristics and life-course profile by trajectory group
| Characteristic | Overall (N = 5,871) |
Stable Marriage & Two-Child (n = 2,349) |
Early Marriage & High-Parity (n = 2,541) |
Stable Marriage & One-Child (n = 913) |
Predominantly Married & Childless (n = 68) |
p-value |
|---|---|---|---|---|---|---|
| Age at baseline (years), mean (SD) | 54.18 (5.91) | 53.06 (5.54) | 56.19 (5.87) | 51.53 (5.00) | 53.26 (6.25) | < 0.001 |
| Birth cohort, n (%) | < 0.001 | |||||
| 1946–1950 | 1031 (17.6%) | 260 (11.1%) | 715 (28.1%) | 48 (5.3%) | 8 (11.8%) | |
| 1951–1955 | 1593 (27.1%) | 601 (25.6%) | 790 (31.1%) | 182 (19.9%) | 20 (29.4%) | |
| 1956–1960 | 1253 (21.3%) | 552 (23.5%) | 457 (18.0%) | 232 (25.4%) | 12 (17.6%) | |
| 1961–1966 | 1994 (34.0%) | 936 (39.8%) | 579 (22.8%) | 451 (49.4%) | 28 (41.2%) | |
| Education, n (%) | < 0.001 | |||||
| Below upper secondary | 5324 (90.7%) | 2152 (91.6%) | 2439 (96.0%) | 667 (73.1%) | 66 (97.1%) | |
| Upper secondary or above | 547 (9.3%) | 197 (8.4%) | 102 (4.0%) | 246 (26.9%) | 2 (2.9%) | |
| Residence, n (%) | < 0.001 | |||||
| Urban | 1187 (20.2%) | 457 (19.5%) | 254 (10.0%) | 463 (50.7%) | 13 (19.1%) | |
| Rural | 4684 (79.8%) | 1892 (80.5%) | 2287 (90.0%) | 450 (49.3%) | 55 (80.9%) | |
| Occupation, n (%) | < 0.001 | |||||
| Agricultural | 2003 (34.1%) | 765 (32.6%) | 851 (33.5%) | 367 (40.2%) | 20 (29.4%) | |
| Non-agricultural | 3868 (65.9%) | 1584 (67.4%) | 1690 (66.5%) | 546 (59.8%) | 48 (70.6%) | |
| Marital status at age 50, n (%) | < 0.001 | |||||
| Married | 5461 (93.0%) | 2224 (94.7%) | 2320 (91.3%) | 856 (93.8%) | 61 (89.7%) | |
| Not married, widowed, or divorced | 410 (7.0%) | 125 (5.3%) | 221 (8.7%) | 57 (6.2%) | 7 (10.3%) | |
| Parity at age 50, n (%) | < 0.001 | |||||
| 0 children | 70 (1.2%) | 0 (0.0%) | 0 (0.0%) | 3 (0.3%) | 67 (98.5%) | |
| 1 child | 901 (15.3%) | 0 (0.0%) | 0 (0.0%) | 900 (98.6%) | 1 (1.5%) | |
| 2 children | 2349 (40.0%) | 2339 (99.6%) | 0 (0.0%) | 10 (1.1%) | 0 (0.0%) | |
| 3 children | 1494 (25.4%) | 9 (0.4%) | 1485 (58.4%) | 0 (0.0%) | 0 (0.0%) | |
| 4 or more children | 1057 (18.0%) | 1 (0.0%) | 1056 (41.6%) | 0 (0.0%) | 0 (0.0%) | |
| Baseline CES-D-10 score (0–30), mean (SD) | 9.01 (6.50) | 8.72 (6.47) | 9.96 (6.63) | 6.87 (5.51) | 8.30 (6.20) | < 0.001 |
| Baseline Nagi score (0–6), mean (SD) | 1.04 (1.29) | 0.97 (1.25) | 1.20 (1.34) | 0.76 (1.15) | 0.73 (0.98) | < 0.001 |
Categorical variables are presented as n (%), and continuous variables as mean (standard deviation). The four trajectory groups were identified using multichannel sequence analysis (MCSA) of marital and fertility histories from ages 15 to 49. P values were derived from one-way analysis of variance for continuous variables and chi-square tests for categorical variables across the four groups
CES-D-10 10-item Center for Epidemiologic Studies Depression Scale, Nagi Nagi Disability Scale
As shown in Table S6, compared to the analytical sample, the excluded participants (N = 217) were slightly younger and had a higher proportion of upper secondary education. While these demographic differences were statistically significant, the small size of the excluded group relative to the final analytical sample suggests that any potential selection bias is likely minimal.
Identification of family life-course trajectories
The family life-course trajectories identified from multichannel sequence analysis are shown in Fig. 2. We selected a four-cluster solution based on complementary clustering quality metrics including an average silhouette width of 0.655, Hubert’s Gamma of 0.963, and Point Biserial Correlation of 0.849 (Table S5).
Fig. 2.
Family life-course trajectories identified by multichannel sequence analysis.The figure displays the four family life-course trajectory groups derived from marital and fertility histories between ages 15 and 49: Stable Marriage & Two-Child, Early Marriage & High-Parity, Stable Marriage & One-Child, and Predominantly Married & Childless. Trajectory groups were identified using multichannel sequence analysis (MCSA) and hierarchical clustering. Cluster proportions and the overall sample size are displayed in the figure. Cluster quality statistics are reported in Table S5
The four trajectory groups were labeled descriptively as Stable Marriage & Two-Child, Early Marriage & High-Parity, Stable Marriage & One-Child, and Predominantly Married & Childless. The labels reflect the dominant features of the state distribution plots rather than formal thresholds for marriage timing or marital stability. The four clusters are characterized as follows:
Stable Marriage & Two-Child (n = 2,349; 40.0%): predominantly stable marriage and two children.
Early Marriage & High-Parity (n = 2,541; 43.3%): early transition to marriage and three or more children.
Stable Marriage & One-Child (n = 913; 15.6%): predominantly stable marriage and one child. Although transitions to marriage and parenthood occurred somewhat later than in the Early Marriage & High-Parity group, the defining feature of this cluster was low parity rather than later marriage timing.
Predominantly Married & Childless (n = 68; 1.2%): near-complete absence of biological children and predominantly married status across adulthood, with somewhat greater marital disruption than the other groups.
The Stable Marriage & Two-Child group was set as the reference group because it was the largest group. The small sample size of the Predominantly Married & Childless group warrants cautious interpretation.
Baseline socioeconomic and health characteristics differed significantly among the four groups (Table 1; all p < .001).
Longitudinal Depression and Functional Limitation Trajectories
The LMM results for depressive symptoms and physical functional limitations are presented in Tables 2 and 3, respectively.
Table 2.
Results of linear mixed-effects models for depressive symptoms (CES-D-10) by life-course fertility and marriage trajectories
| Variable | β (SE) | 95% CI | p-value |
|---|---|---|---|
| Intercept | 7.928 (0.243)*** | [7.452, 8.405] | < 0.001 |
| Trajectory group (ref: Stable Marriage & Two-Child) | |||
| Stable Marriage & One-Child | -0.858 (0.211)*** | [-1.271, -0.445] | < 0.001 |
| Early Marriage & High-Parity | 0.858 (0.152)*** | [0.560, 1.157] | < 0.001 |
| Predominantly Married & Childless | 0.804 (0.652) | [-0.473, 2.082] | 0.217 |
| Time items | |||
| Time (linear, years) | 0.078 (0.022)*** | [0.034, 0.122] | < 0.001 |
| Time squared | 0.037 (0.006)*** | [0.026, 0.049] | < 0.001 |
| Covariates | |||
| Upper secondary or above (ref: Below upper secondary) | -1.474 (0.253)*** | [-1.970, -0.978] | < 0.001 |
| Rural (ref: Urban) | 1.725 (0.195)*** | [1.343, 2.107] | < 0.001 |
| Non-agricultural (ref: Agricultural) | -0.403 (0.150)** | [-0.698, -0.109] | 0.007 |
| Birth cohort (ref: 1946–1950) | |||
| 1951–1955 | -0.240 (0.207) | [-0.646, 0.165] | 0.245 |
| 1956–1960 | -0.332 (0.223) | [-0.769, 0.105] | 0.137 |
| 1961–1966 | -0.707 (0.211)*** | [-1.120, -0.293] | < 0.001 |
| Trajectory × Time interaction | |||
| Stable Marriage & One-Child × Time | 0.012 (0.043) | [-0.071, 0.096] | 0.772 |
| Early Marriage & High-Parity × Time | 0.008 (0.031) | [-0.052, 0.068] | 0.797 |
| Predominantly Married & Childless × Time | 0.231 (0.156) | [-0.074, 0.536] | 0.138 |
β (SE) and 95% confidence intervals are reported. The dependent variable was the CES-D-10 score (range 0–30), with higher scores indicating more severe depressive symptoms. N = 5,827. Time was centered at the midpoint of follow-up (3.25 years), so the intercept and trajectory main effects represent estimated values at the centered reference time point. A quadratic time term (Time²) was included to allow for nonlinear change over time. Models were adjusted for education, residence, occupation, and birth cohort. *** p < .001, ** p < .01, * p < .05
Table 3.
Results of linear mixed-effects models for physical functional limitations (Nagi) by life-course fertility and marriage trajectories
| Variable | β (SE) | 95% CI | p-value |
|---|---|---|---|
| Intercept | 1.481 (0.050)*** | [1.383, 1.579] | < 0.001 |
| Trajectory group (ref: Stable Marriage & Two-Child) | |||
| Stable Marriage & One-Child | -0.163 (0.045)*** | [-0.251, -0.076] | < 0.001 |
| Early Marriage & High-Parity | 0.206 (0.032)*** | [0.143, 0.269] | < 0.001 |
| Predominantly Married & Childless | 0.065 (0.138) | [-0.205, 0.335] | 0.636 |
| Time items | |||
| Time (linear, years) | 0.067 (0.005)*** | [0.057, 0.076] | < 0.001 |
| Covariates | |||
| Upper secondary or above (ref: Below upper secondary) | -0.191 (0.053)*** | [-0.294, -0.088] | < 0.001 |
| Rural (ref: Urban) | 0.317 (0.041)*** | [0.238, 0.396] | < 0.001 |
| Non-agricultural (ref: Agricultural) | -0.411 (0.031)*** | [-0.472, -0.350] | < 0.001 |
| Birth cohort (ref: 1946–1950) | |||
| 1951–1955 | -0.145 (0.043)*** | [-0.229, -0.061] | < 0.001 |
| 1956–1960 | -0.212 (0.046)*** | [-0.303, -0.121] | < 0.001 |
| 1961–1966 | -0.429 (0.044)*** | [-0.514, -0.343] | < 0.001 |
| Trajectory × Time interaction | |||
| Stable Marriage & One-Child × Time | -0.018 (0.009) † | [-0.036, 0.000] | 0.053 |
| Early Marriage & High-Parity × Time | 0.025 (0.007)*** | [0.012, 0.038] | < 0.001 |
| Predominantly Married & Childless × Time | 0.058 (0.033) † | [-0.006, 0.122] | 0.078 |
β (SE) and 95% confidence intervals are reported. The dependent variable was the Nagi score (range 0–6), with higher scores indicating more self-reported physical functional limitations. N = 5,870. The physical functional limitations model included a linear time term. Time was centered at the midpoint of follow-up (3.25 years), so the intercept and trajectory main effects represent estimated values at the centered reference time point. Models were adjusted for education, residence, occupation, and birth cohort. *** p < .001, ** p < .01, * p < .05, †p < .10
For depressive symptoms, women in the Early Marriage & High-Parity group had higher symptom levels than the reference group (β = 0.858, 95% CI: 0.560 to 1.157, p < .001), whereas women in the Stable Marriage & One-Child group had lower symptom levels (β = -0.858, 95% CI: -1.271 to -0.445, p < .001). The Predominantly Married & Childless group did not differ significantly from the reference group at the centered reference time point. No significant between-group differences in the rate of change in depressive symptoms were observed over follow-up (Fig. 3). For physical functional limitations, the trajectories diverged over time. Compared with the reference group, the Early Marriage & High-Parity group showed a significantly faster increase in functional limitations (trajectory × time interaction β = 0.025, 95% CI: 0.012 to 0.038, p < .001). The Stable Marriage & One-Child group showed a marginally slower increase in functional limitations (trajectory × time interaction β = -0.018, 95% CI: -0.036 to 0.000, p = .053). The Predominantly Married & Childless group showed a numerically larger increase in functional limitations (trajectory × time interaction β = 0.058, 95% CI: -0.006 to 0.122, p = .078), though this did not reach conventional statistical significance (Fig. 4).
Fig. 3.
Predicted longitudinal trajectories of depressive symptoms by family life-course group. Predicted values are derived from the primary linear mixed-effects model with linear and quadratic time terms for depressive symptoms, adjusted for education, residence, occupation, and birth cohort. The x-axis labels represent the actual survey years (2011–2018), whereas the model used time centered at the follow-up midpoint (3.25 years). Shaded areas denote 95% confidence intervals (CIs). Total analytic sample: N = 5,827
Fig. 4.
Predicted longitudinal trajectories of physical functional limitations by family life-course group. Predicted values are derived from the primary linear mixed-effects model for physical functional limitations, adjusted for education, residence, occupation, and birth cohort. The x-axis labels represent the actual survey years (2011–2018), whereas the model used time centered at the follow-up midpoint (3.25 years). The y-axis is scaled to show predicted Nagi scores from 0.8 to 2.2. Shaded areas denote 95% confidence intervals (CIs). Total analytic sample: N = 5,870
Exploratory analysis of intergenerational support and social participation
Exploratory analyses revealed significant differences in intergenerational support and social participation across family life-course trajectory groups (Table S1). Women in the Stable Marriage & One-Child group received significantly less financial support from their children than those in the Stable Marriage & Two-Child and Early Marriage & High-Parity groups (9.9% vs. 25.1% and 39.9%, respectively; p < .001; adjusted OR = 0.44, 95% CI: 0.34–0.58). However, this group exhibited the highest rates of social participation (61.8%) and engagement in dancing or fitness activities (18.0%; adjusted OR = 1.82, 95% CI: 1.34–2.47). In contrast, the Early Marriage & High-Parity group had the highest proportion of financial support but the lowest participation in dancing or fitness activities (3.1%, p < .001).
There were also significant differences in the number of types of social activities participated in among the trajectory groups, with the Stable Marriage & One-Child group engaging in a greater variety of activities on average, and the Early Marriage & High-Parity group participating in fewer. No significant differences were observed between groups in co-residence with children, caregiving support, or weekly contact frequency.
Robustness analysis adjusted for childhood contexts
After further adjustment for childhood contexts before age 15, the main associations remained robust (Table S2). Compared with the Stable Marriage & Two-Child group, the Stable Marriage & One-Child group remained associated with lower depressive symptoms (β = -0.843, 95% CI: -1.247 to -0.439, p < .001) and lower physical functional limitations (β = -0.162, 95% CI: -0.249 to -0.076, p < .001), while the Early Marriage & High-Parity group remained associated with higher depressive symptoms (β = 0.800, 95% CI: 0.508 to 1.093, p < .001) and higher physical functional limitations (β = 0.196, 95% CI: 0.134 to 0.259, p < .001). The Predominantly Married & Childless group remained non-significant for both outcomes.
For physical functional limitations, the Stable Marriage & One-Child group still showed a slower increase over time (trajectory × time: β = -0.019, 95% CI: -0.037 to -0.001, p = .039), whereas the Early Marriage & High-Parity group continued to show a faster increase (trajectory × time: β = 0.024, 95% CI: 0.011 to 0.037, p < .001). The trajectory-by-time interactions for depressive symptoms remained non-significant.
Childhood famine exposure, poorer childhood health, and poorer childhood financial status were each associated with higher depressive symptoms and greater physical functional limitations (all p < .001). Overall, the direction and significance of the primary associations were unchanged after accounting for childhood disadvantage.
Robustness analyses of trajectory-by-time interactions
Trajectory-by-time interaction results were generally consistent across sensitivity analyses (Table S3). For depressive symptoms, no clear trajectory-by-time differences were observed after additional adjustment for childhood adversity or after restricting the sample to women born between 1951 and 1966. Only after excluding participants with baseline CES-D-10 scores greater than 0 did the interaction term for the Stable Marriage & One-Child group become significantly negative (β = -0.424, 95% CI: -0.711 to -0.138, p < .01).
For physical functional limitations, additional adjustment for childhood adversity rendered the interaction term for the Stable Marriage & One-Child group statistically significant (β = -0.019, 95% CI: -0.037 to -0.001, p = .039), while the Early Marriage & High-Parity group remained significantly positive (β = 0.024, 95% CI: 0.011 to 0.037, p < .001). In the baseline Nagi = 0 only analysis, which addressed the concentration of scores at the floor of the scale, the direction of the interactions remained the same: the Stable Marriage & One-Child group showed a slower increase in functional limitations (β = -0.027, SE = 0.012, p < .05), and the Early Marriage & High-Parity group showed a faster increase (β = 0.033, SE = 0.009, p < .001). These findings suggest that the trajectory differences in functional limitations were not solely driven by pre-existing baseline limitations. A similar pattern was observed when the sample was restricted to women born between 1951 and 1966.
Sensitivity analysis of participants with ≥ 2 valid outcome measurements and sample comparison
Sensitivity analyses restricted to participants with at least two valid outcome observations yielded results that were broadly consistent with the main analysis (Table S4), with the overall ranking of trajectory groups and the direction of the associations remaining essentially unchanged. The Stable Marriage & One-Child group continued to show lower levels, whereas the Early Marriage & High-Parity group continued to show higher levels, relative to the Stable Marriage & Two-Child reference group.
Included and excluded participants differed in several baseline characteristics (Table S6). Compared with included participants, excluded participants were slightly younger (52.71 vs. 54.18 years, p < .001), more likely to have attained upper secondary education or above (17.1% vs. 9.3%, p < .001), and more concentrated in the younger birth cohort (1961–1966: 42.4% vs. 34.0%, p = .006). Occupation also differed: excluded participants were slightly more likely to be in agricultural occupations (35.5% vs. 34.1%) and less likely to be in non-agricultural occupations (49.8% vs. 65.9%), with a higher proportion of missing occupation data (14.7%, p = .042). Because several variables had substantial missingness among excluded participants, these comparisons should be interpreted cautiously.
Discussion
This 7-year longitudinal study identified distinct family life-course trajectories among Chinese women and showed that these trajectories were associated with subsequent changes in depressive symptoms and physical functional limitations. Compared with the Stable Marriage & Two-Child group, the Early Marriage & High-Parity group had higher baseline depressive symptoms and greater functional limitations, and experienced a faster increase in functional limitations over time. By contrast, the Stable Marriage & One-Child group had lower depressive symptoms and lower functional limitations, and tended to experience a slower increase in functional limitations. The Predominantly Married & Childless group did not differ significantly from the reference group. Importantly, the observed differences were modest in magnitude and reflected gradual divergence over time rather than abrupt separation of trajectories.
These findings are broadly consistent with the life-course literature, which suggests that marital and reproductive histories should be understood as cumulative patterns rather than isolated events. They also add evidence from China, where family formation has been shaped by strong kinship norms, marked socioeconomic stratification, and changing fertility policy. In this respect, cross-national studies remind us that the meaning of family trajectories is context-dependent. For instance, while our study suggests health disadvantages among women with early marriage and high parity, such associations were not found in recent evidence from France, where divorce showed minimal impact on late-life well-being [37]. Similarly, research in the United Kingdom suggests that the health effects of family trajectories are highly cohort-sensitive [38]. These contrasts underscore that family trajectories are not universal stressors; instead, their impact is filtered through China’s specific gender norms, welfare arrangements, and policy environments. In the Chinese context, where formal social protection was limited for the 1946–1966 cohorts, family configurations remained a primary determinant of resource access, whereas in contexts like France, institutional safety nets may buffer the health risks associated with family transitions [37].
The poorer outcomes observed in the Early Marriage & High-Parity group are consistent with, but do not prove, the maternal depletion perspective [5, 39]. Higher parity may entail longer exposure to pregnancy and childrearing demands, but it also reflects restricted professional pathways and economic accumulation [40]. Conversely, the physical functional limitations advantage of the Stable Marriage & One-Child group should be interpreted through both biological and sociological lenses. While fewer births may reduce reproductive burden, our exploratory analysis (Table S1) suggests that this group had more favorable social participation profiles, reporting the highest rates of social participation (61.8%) and fitness activities (18.0%) [41]. As sociological frameworks suggest, these women may have followed more advantaged occupational and social pathways, which could have contributed to greater economic autonomy and fewer cumulative work-family strains than those experienced by high-parity mothers. Thus, their slower functional decline may reflect a combination of reduced reproductive burden and a greater accumulation of socioeconomic buffers over the life course. The Predominantly Married & Childless group did not show significant differences, which may partly reflect its small size and the heterogeneity of childlessness in this cohort [42].
Beyond family trajectories themselves, childhood adversity appeared to remain relevant. Although additional adjustment for childhood famine, childhood health, and childhood financial status did not materially change the main associations, these early-life factors were themselves related to poorer depressive and functional outcomes [29]. This suggests that family trajectories should be viewed in the context of earlier life disadvantage, rather than as isolated exposures. More broadly, the results support a cumulative life-course perspective in which early conditions, family formation, and later health are linked through layered social processes.
The exploratory analysis of intergenerational support and social participation offers further context. The Stable Marriage & One-Child group received less financial support from children but had higher social participation, particularly in dancing or fitness activities. In contrast, the Early Marriage & High-Parity group was more likely to receive financial support but participated less in social activities. These patterns suggest that support is not uniform across trajectories and that family structure may be related to later health through both kin-based resources and broader social engagement. The analysis was descriptive, but it points to potentially important pathways that deserve further investigation.
Using multichannel sequence analysis enabled us to capture the timing, order, and duration of marital and fertility experiences in ways that single-point indicators of parity or marital status would not have permitted [22, 26]. This is important because the health differences observed here would likely have been missed with a simpler measure of marital status or parity. At the same time, the relatively modest effect sizes indicate that family trajectories are only one part of a broader set of influences on later health. In this sample, which spans midlife to early old age, the associations appear to reflect gradual accumulation of disadvantage rather than dramatic health shocks [43].
The Chinese context also matters for interpreting these results. Although formal social protection has expanded, family support remains central for many women, and the availability of long-term care is still uneven [44]. Family trajectories may therefore influence health not only through reproductive history, but also through later access to practical support, social participation, and care arrangements. This may help explain why the observed patterns are more evident for physical functional limitations than for depressive symptoms, and why the differences remain limited in size.
Several limitations should be noted. First, the Early Marriage & High-Parity trajectory should not be interpreted as a direct policy effect, as it reflects both cohort conditions and broader socioeconomic change. Second, although we adjusted for education, residence, occupation, birth cohort, and childhood adversity, residual confounding cannot be excluded [45]. Third, the sample covers women aged 45–65 at baseline and followed them to approximately 52–72 years, so the findings are more relevant to the transition from midlife to early old age than to advanced old age. Fourth, although the Nagi scale is widely used to assess functional limitations, it should not be interpreted as a direct measure of intrinsic physical functioning. The score was concentrated at the lower end of the scale, with 47.1% of women reporting no functional limitations at baseline, which may have reduced variability and limited the ability to detect small between-group differences. However, the sensitivity analysis restricted to women with baseline Nagi scores of 0 yielded a similar pattern of trajectory-by-time interactions, suggesting that the observed functional limitation trajectories were not solely driven by baseline functional status. Finally, despite multiple sensitivity analyses, attrition may still have introduced some bias if the most vulnerable participants were more likely to be lost to follow-up [30].
In summary, women’s marital and fertility trajectories were associated with later changes in depressive symptoms and physical functional limitations, but the differences were modest and developed gradually over time. These associations should be interpreted in relation to the early-life conditions and social inequalities that family trajectories both reflect and reinforce [46].
Supplementary Information
Supplementary Material 1. Table S1. Exploratory Analysis of Intergenerational Support and Social Participation. Table S2. Robustness Analysis Adjusted for Childhood Contexts Before Age 15. Table S3. Robustness Analyses of Trajectory-by-Time Interaction Effects. Table S4. Sensitivity Analysis: Restricting to Participants with ≥2 Valid Outcome Measurements. Table S5. Cluster Quality Statistics for Multichannel Sequence Analysis. Table S6. Comparison of Baseline Characteristics Between Included and Excluded Participants.
Acknowledgements
We thank the China Center for Economic Research and the National School of Development at Peking University for providing access to the Harmonized China Health and Retirement Longitudinal Study (CHARLS) data. We also thank the CHARLS research team and all participants for their contribution to the data collection.
Abbreviations
- CHARLS
China Health and Retirement Longitudinal Study
- MCSA
Multichannel Sequence Analysis
- LMM
Linear Mixed-Effects Model
- CES-D-10
10-item Center for Epidemiologic Studies Depression Scale
- Nagi
Nagi Disability Scale
- OR
Odds Ratio
- CI
Confidence Interval
Authors' contributions
Cheng-Chuan Chen: Conceptualization, Methodology, Writing - Original Draft. Na Zhou: Data curation, Formal analysis, Software. Xiao-Wen Wu: Visualization, Validation, Writing - Original Draft. Yun Qiao: Investigation, Resources. Xiao-Bin Wang: Supervision, Project administration, Funding acquisition, Writing - Review & Editing.
Funding
This work was supported by the Natural Science Foundation of Sichuan Province (Grant No. 2024NSFSC2096) and the Joint Project of Luzhou City and Southwest Medical University (Grant No. 2025LZXNYDYG05). The funders had no role in study design, data collection, analysis, interpretation, or manuscript preparation.
Data availability
The data that support the findings of this study are openly available from the CHARLS project website (http://charls.pku.edu.cn/en) upon registration.
Declarations
Ethics approval and consent to participate
The original CHARLS study was approved by the Ethical Review Committee of Peking University (Approval No. IRB00001052-11015). All procedures were conducted in accordance with the Declaration of Helsinki, and all participants provided written informed consent at the time of original participation. Because this study involved the secondary analysis of publicly available, de-identified data, further ethical review was not required by the authors’ institutional review boards.
Consent for publication
Not applicable.
Competing interests
The authors declare no competing interests.
Footnotes
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Cheng-Chuan Chen, Na Zhou and Xiao-Wen Wu contributed equally to this work.
References
- 1.Lesthaeghe R. The unfolding story of the second demographic transition. Popul Dev Rev. 2010;36(2):211–51. [DOI] [PubMed] [Google Scholar]
- 2.Grundy E, Read S. Pathways from fertility history to later life health: Results from analyses of the English Longitudinal Study of Ageing. Demographic Res. 2015;32:107–46. [Google Scholar]
- 3.Silverstein M, Conroy SJ, Wang H, Giarrusso R, Bengtson VL. Reciprocity in parent–child relations over the adult life course. Journals Gerontol Ser B: Psychol Sci Social Sci. 2002;57(1):S3–13. [DOI] [PubMed] [Google Scholar]
- 4.Kuh D, Ben-Shlomo Y, Lynch J, Hallqvist J, Power C. Life course epidemiology. J Epidemiol Commun Health. 2003;57(10):778–83. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.King JC. The risk of maternal nutritional depletion and poor outcomes increases in early or closely spaced pregnancies. J Nutr. 2003;133(5):S1732–6. [DOI] [PubMed] [Google Scholar]
- 6.Winkvist A, Rasmussen KM, Habicht JP. A new definition of maternal depletion syndrome. Am J Public Health. 1992;82(5):691–4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Mollborn S. Teenage mothers today: What we know and how it matters. Child Dev Perspect. 2017;11(1):63–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Miller AR. The effects of motherhood timing on career path. J Popul Econ. 2011;24(3):1071–100. [Google Scholar]
- 9.Spence NJ. The long-term consequences of childbearing: Physical and psychological well-being of mothers in later life. Res Aging. 2008;30(6):722–51. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Berkman LF, Glass T, Brissette I, Seeman T. E. From social integration to health: Durkheim in the new millennium. Soc Sci Med. 2000;51(6):843–57. [DOI] [PubMed] [Google Scholar]
- 11.Umberson D, Williams K, Powers DA, Liu H, Needham B. You make me sick: Marital quality and health over the life course. J Health Soc Behav. 2006;47(1):1–16. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Chen F, Short SE. Household context and subjective well-being among the oldest old in China. J Fam Issues. 2008;29(10):1379–403. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Whyte MK. Filial obligations in Chinese families: Paradoxes of modernization. In: Ikels C, editor. Filial piety: Practice and discourse in contemporary East Asia. Stanford University Press; 2004. pp. 106–27.
- 14.Zimmer Z, Kwong J. Family size and support of older adults in urban and rural China: Current effects and future implications. Demography. 2003;40(1):23–44. [DOI] [PubMed] [Google Scholar]
- 15.Poston DL. Social and economic development and the fertility transitions in mainland China and Taiwan. Popul Dev Rev. 2000;26(Suppl):40–60. [Google Scholar]
- 16.Short SE, Zhai F. Looking locally at China’s one-child policy. Stud Fam Plann. 1998;29(4):373–87. [PubMed] [Google Scholar]
- 17.Ferraro KF, Shippee TP. Aging and cumulative inequality: How does inequality get under the skin? Gerontologist. 2009;49(3):333–43. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.O’Rand AM. The precious and the precocious: Understanding cumulative disadvantage and cumulative advantage over the life course. Gerontologist. 1996;36(2):230–8. [DOI] [PubMed] [Google Scholar]
- 19.Zhang C, Lei X, Strauss J, Zhao Y. Health insurance and health care among the mid-aged and older Chinese: Evidence from the national baseline survey of CHARLS. Health Econ. 2017;26(4):431–49. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Chen X, Giles J, Yao Y, Yip W, Meng Q, Berkman L, Wang F. The path to healthy ageing in China: A Peking University–Lancet Commission. Lancet. 2022;400(10367):1967–2006. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Raymo JM, Park H, Xie Y, Yeung WJJ. Marriage and family in East Asia: Continuity and change. Ann Rev Sociol. 2015;41:471–92. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Aisenbrey S, Fasang AE. New life for old ideas: The second wave of sequence analysis bringing the course back into the life course. Sociological Methods & Res; 2010;38:420–62. [Google Scholar]
- 23.Elder GH. Jr. The life course as developmental theory. Child Dev. 1998;69(1):1–12. [PubMed] [Google Scholar]
- 24.Wu J, Marois G. Family-work history and inequalities in old-age cognition in China. Soc Sci Med. 2025;381:118318. 10.1016/j.socscimed.2025.118318. [DOI] [PubMed] [Google Scholar]
- 25.Hongzhou C. Work-family life course and depressive symptoms in later life among Chinese middle-age and older adults. Adv Life Course Res. 2026;100729. 10.1016/j.alcr.2026.100729. [DOI] [PubMed]
- 26.Chen J, Vidal S, Spijker J. Work-family trajectories and later-life health in China. Adv Life Course Res. 2026;100740. 10.1016/j.alcr.2026.100740. [DOI] [PubMed]
- 27.Lei X, Sun X, Strauss J, Zhang P, Zhao Y. Depressive symptoms and SES among the mid-aged and elderly in China: Evidence from the China Health and Retirement Longitudinal Study national baseline. Soc Sci Med. 2014;120:224–32. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Beydoun MA, Popkin BM. The impact of socio-economic factors on functional status decline among community-dwelling older adults in China. Soc Sci Med. 2005;60(9):2045–57. [DOI] [PubMed] [Google Scholar]
- 29.Haas SA. Trajectories of functional health: The long arm of childhood health and socioeconomic factors. Soc Sci Med. 2008;66(4):849–61. [DOI] [PubMed] [Google Scholar]
- 30.Little RJA. Modeling the drop-out mechanism in repeated-measures studies. J Am Stat Assoc. 1995;90(431):1112–21. [Google Scholar]
- 31.Zhao Y, Hu Y, Smith JP, Strauss J, Yang G. Cohort profile: The China Health and Retirement Longitudinal Study (CHARLS). Int J Epidemiol. 2014;43(1):61–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Gabadinho A, Ritschard G, Müller NS, Studer M. Analyzing and visualizing state sequences in R with TraMineR. J Stat Softw. 2011;40(4):1–37. [Google Scholar]
- 33.Chen Y, Zhou LA. The long-term health and economic consequences of the 1959–1961 famine in China. J Health Econ. 2007;26(4):659–81. [DOI] [PubMed] [Google Scholar]
- 34.Smith JP. Reconstructing childhood health histories. Demography. 2009;46(2):387–403. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Raudenbush SW, Bryk AS. Hierarchical linear models: Applications and data analysis methods. 2nd ed. Sage; 2002. [Google Scholar]
- 36.Glymour MM, Weuve J, Berkman LF, Kawachi I, Robins JM. When is baseline adjustment useful in analyses of change? An example with education and cognitive change. Am J Epidemiol. 2005;162(3):267–78. [DOI] [PubMed] [Google Scholar]
- 37.Beaufils C, Barbuscia A, Cambois E. Women’s employment–family trajectories and well-being in later life: Evidence from France. J Aging Health. 2023;35(7–8):577–92. [DOI] [PubMed] [Google Scholar]
- 38.Lacey RE, McMunn A, Webb E. Family careers and health during midlife: A life course study of a British birth cohort. Longitud Life Course Stud. 2017;8(2):143–59. [Google Scholar]
- 39.Grundy E, Kravdal Ø. Fertility history and cause-specific mortality: A register-based analysis of complete cohorts of Norwegian women and men. Soc Sci Med. 2010;70(11):1847–57. [DOI] [PubMed] [Google Scholar]
- 40.Budig MJ, England P. The wage penalty for motherhood. Am Sociol Rev. 2001;66(2):204–25. [Google Scholar]
- 41.Holt-Lunstad J, Smith TB, Layton JB. Social relationships and mortality risk: A meta-analytic review. PLoS Med. 2010;7(7),e1000316. [DOI] [PMC free article] [PubMed]
- 42.Umberson D, Pudrovska T, Reczek C. Parenthood, childlessness, and well-being: A life course perspective. J Marriage Family. 2010;72(3):612–29. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Dannefer D. Cumulative advantage/disadvantage and the life course: Cross-fertilizing age and social science theory. The Journals of Gerontology Series B: Psychological Sciences and Social Sciences. 2003;58(6):S327–S337. [DOI] [PubMed]
- 44.Feng Z, Liu C, Guan X, Mor V. China’s rapidly aging population creates policy challenges in shaping a viable long-term care system. Health Aff. 2012;31(12):2764–73. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.Fewell Z, Davey Smith G, Sterne JAC. The impact of residual and unmeasured confounding in epidemiologic studies: A simulation study. Am J Epidemiol. 2007;166(6):646–55. [DOI] [PubMed] [Google Scholar]
- 46.Ben-Shlomo Y, Kuh D. A life course approach to chronic disease epidemiology: Conceptual models, empirical challenges and interdisciplinary perspectives. Int J Epidemiol. 2002;31(2):285–93. [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Supplementary Material 1. Table S1. Exploratory Analysis of Intergenerational Support and Social Participation. Table S2. Robustness Analysis Adjusted for Childhood Contexts Before Age 15. Table S3. Robustness Analyses of Trajectory-by-Time Interaction Effects. Table S4. Sensitivity Analysis: Restricting to Participants with ≥2 Valid Outcome Measurements. Table S5. Cluster Quality Statistics for Multichannel Sequence Analysis. Table S6. Comparison of Baseline Characteristics Between Included and Excluded Participants.
Data Availability Statement
The data that support the findings of this study are openly available from the CHARLS project website (http://charls.pku.edu.cn/en) upon registration.