Abstract
Background
Arthritis represents a major global health burden, with social economic status (SES) emerging as a critical determinant. However, longitudinal evidence from Asian populations remains limited.
Methods
This prospective cohort study used China Health and Retirement Longitudinal Study (CHARLS) data, including 4,469 arthritis-free participants aged ≥ 45 years from baseline (2011) through 2018. SES was assessed using composite scores combining education and household wealth, categorized into four groups. New-onset arthritis was determined by self-reported physician diagnosis. Covariates included demographics, lifestyle factors, anthropometrics, and comorbidities. Multivariable logistic and Cox regression models estimated odds ratios (ORs) and hazard ratios (HRs) with 95% confidence intervals (CIs). Kaplan-Meier analysis estimated cumulative incidence with log-rank tests. Subgroup analyses examined effect modification by gender and age.
Results
During 7-year follow-up, 1,324 participants (29.6%) developed arthritis. In fully adjusted models, compared with low SES, upper-middle SES demonstrated reduced risk in both logistic (OR = 0.67, 95% CI: 0.55–0.82, P < 0.001) and Cox regression (HR = 0.71, 95% CI: 0.60–0.84, P < 0.001). High SES showed greater protection (OR = 0.41, 95% CI: 0.14-1.00; HR = 0.44, 95% CI: 0.18–1.08). Kaplan-Meier analysis showed cumulative incidence decreased across SES groups: 34.5% (low), 31.1% (low-middle), 23.5% (upper-middle), and 14.3% (high SES; log-rank P < 0.0001). Effects remained consistent across gender and age subgroups.
Conclusions
Higher SES was independently associated with reduced arthritis risk among Chinese middle-aged and elderly adults, demonstrating a clear dose-response relationship. These findings support targeted prevention strategies for socioeconomically disadvantaged populations.
Supplementary Information
The online version contains supplementary material available at 10.1186/s12889-025-24569-0.
Keywords: Social economic status, Arthritis, Longitudinal study, Middle-aged and elderly, CHARLS
Introduction
Arthritis represents a substantial health burden among middle-aged and elderly populations globally, with its incidence showing a persistent upward trend. According to a systematic analysis for the Global Burden of Disease Study in 2021, 595 million people globally had osteoarthritis in 2020, affecting 7.6% of the global population and making it the most common form of arthritis, with particularly significant impact on populations over 30 years of age, affecting 15% of this demographic [1]. In China, the epidemiological characteristics of arthritis are particularly pronounced due to accelerated population aging. Recent epidemiological surveys indicate that the prevalence of osteoarthritis among Chinese individuals aged 65 and older is 8.1% [2]. The Global Burden of Disease (GBD) 2019 study demonstrates that osteoarthritis has become a leading cause of disability-adjusted life years (DALYs), imposing enormous socioeconomic burdens worldwide [3]. Particularly in low- and middle-income countries, where two-thirds of the population aged 60 and older are projected to reside by 2050, the disease burden of arthritis is expected to intensify further [4].
Social economic status (SES), as a critical social determinant of health outcomes, typically encompasses a comprehensive assessment of multiple dimensions including educational attainment, income level, and occupational status [5, 6]. SES influences health through multiple mechanisms: lower SES is often associated with limited healthcare access, adverse lifestyle behaviors, chronic stress exposure, and poor residential environment quality [7, 8]. Extensive research in recent years has confirmed the close relationship between SES and the incidence and adverse outcomes of various chronic diseases. A systematic review published in 2023 revealed that SES changes significantly impact health, with consistently high SES populations demonstrating optimal health outcomes [9]. In cardiovascular disease and cancer, populations with low SES exhibit higher incidence rates, poorer treatment adherence, and elevated mortality rates [10, 11]. Furthermore, SES disparities are closely associated with healthcare utilization inequalities, health behavior differences, and variations in social support networks [12, 13].
Existing research on the relationship between SES and arthritis presents a complex landscape. Some studies suggest that low SES is associated with increased arthritis prevalence, potentially due to factors such as high physical labor intensity, poor nutritional status, and limited healthcare access [14, 15]. However, other study have found that the association between SES and arthritis varies across different populations and geographic regions [16]. Moreover, existing studies exhibit considerable heterogeneity in SES measurement methods, arthritis definitions, follow-up duration, and covariate adjustments, limiting the comparability and generalizability of findings. Particularly in Asian populations, longitudinal research data remain limited, necessitating more high-quality cohort studies to elucidate this association.
Given the limitations and knowledge gaps in existing research, particularly the lack of evidence from large-scale longitudinal cohorts in Asian populations, this study aims to prospectively evaluate the association between SES and new-onset arthritis using data from the China Health and Retirement Longitudinal Study (CHARLS). Our investigation offers several methodological advantages over previous research: (1) a large sample size of 4,469 arthritis-free middle-aged and elderly participants, representing one of the largest cohorts in Asian populations; (2) an extended 7-year follow-up period, longer than most existing longitudinal studies in this field; and (3) comprehensive adjustment for multiple categories of confounding factors including demographic characteristics, lifestyle behaviors, anthropometric measures, and clinical comorbidities, enabling robust causal inference. This study will provide important epidemiological evidence for understanding the role of socioeconomic factors in arthritis development and offer a scientific basis for developing population-based arthritis prevention strategies.
Methods
Ethical considerations
The CHARLS project obtained approval from the Biomedical Ethics Review Committee at Peking University (IRB00001052-11015) [17]. All study participants provided written consent prior to data collection. Research protocols followed international ethical guidelines, including the Declaration of Helsinki. Anonymized data are made publicly accessible via the official study repository (http://charls.pku.edu.cn/).
Study participants
This study utilized data from the CHARLS, a nationally representative prospective cohort investigating health and socioeconomic factors among Chinese adults aged ≥ 45 years, with baseline data collected in 2011 and follow-up assessments conducted in 2013, 2015, and 2018 using a multistage probability-proportional-to-size sampling strategy across 150 counties/districts in 28 provinces throughout China. The CHARLS baseline survey (2011) achieved a household response rate of 80.5% and individual response rate of 80.1%. Follow-up waves maintained high retention rates: 88.8% (2013), 85.9% (2015), and 82.4% (2018). From the 2011 initial baseline sample of 17,705 participants, we systematically excluded individuals with missing SES data (n = 5,222), baseline arthritis (n = 4,143) or missing arthritis information (n = 76), age < 45 years (n = 349) or missing age data (n = 76), incomplete covariate information (n = 1,568), and those lacking arthritis follow-up data during 2013–2018 waves (n = 1,848), resulting in a final analytical cohort of 4,469 participants. This represents a follow-up completion rate of 71.8% among participants eligible for analysis (4,469/6,225 after excluding baseline arthritis cases). Data flow chart and collectiots provided as Fig. 1 and supplementary materials.
Fig. 1.
Participant selection flowchart. Notes: Flowchart showing the selection of participants from the 2011 CHARLS baseline sample (n = 17,705) to the final analytical cohort (n = 4,469)
Assessment of SES
SES was assessed using a composite measure encompassing educational level and total household wealth, following the established and validated methodology used in multiple high-impact studies [18, 19]. This standardized approach has been proven effective for capturing socioeconomic disparities in health outcomes among middle-aged and elderly populations. Educational attainment was standardized into three categories: primary education (scored 0), secondary education including vocational training (scored 1), and tertiary education (scored 2). This hierarchical scoring reflects the progressive nature of educational advantages in health-related knowledge, healthcare access, and health behaviors. Total household wealth was calculated as the sum of all wealth components including residence, business assets, vehicles, and savings accounts, excluding debts and loans. Household wealth was subsequently divided into quartiles within the study population, with scores ranging from 0 (lowest quartile, Q1) to 3 (highest quartile, Q4), ensuring equal population distribution while capturing wealth-related health advantages. The overall SES score was derived by summing educational level and household wealth scores with equal weighting, resulting in a composite score ranging from 0 to 5.
Assessment of arthritis
New-onset arthritis was ascertained based on self-reported physician diagnosis during follow-up waves conducted in 2013, 2015, and 2018. At each assessment, participants were asked “Have you been diagnosed with arthritis by a doctor?” with affirmative responses defining incident arthritis cases. Participants with baseline arthritis (2011) were excluded to ensure a disease-free cohort at study entry. Upon first positive response during follow-up, participants were classified as having new-onset arthritis and censored from further observation to establish precise disease onset timing. Arthritis-free individuals continued follow-up until study completion or loss to follow-up.
Assessment of covariates
Comprehensive baseline data (2011) were collected across demographic, social, lifestyle, anthropometric, and clinical dimensions. Demographic variables comprised age (continuous) and sex (male/female). Social characteristics included marital status, categorized as marital status (including those living separately) or single (partnered, separated, divorced, widowed, never married), and residential setting (urban/rural). Lifestyle assessments captured Smoking history and drinking history as binary variables. Anthropometric measurements utilized Body Mass Index (kg/m²), stratified according to Chinese population criteria [20]: underweight (≤ 18.5), normal (18.5–24.0), overweight (24.0–28.0), and obese (> 28.0). Clinical variables included physician-diagnosed hypertension and diabetes mellitus. All information was obtained through validated questionnaires administered via structured interviews by certified researchers following standardized methodology and rigorous quality control procedures.
Statistical analysis
Baseline characteristics were compared across SES groups using appropriate statistical tests. Continuous variables were presented as means ± standard deviations and compared using one-way analysis of variance (ANOVA). Categorical variables were expressed as frequencies and percentages, with between-group differences assessed using Pearson’s chi-square test of independence. Multivariable logistic regression models were employed to calculate odds ratios (ORs) and 95% confidence intervals (CIs). Cox proportional hazards regression models were used to estimate hazard ratios (HRs) and 95% CIs for the risk of arthritis during follow-up. For both analyses, we constructed three sequential models: Model 1 (unadjusted); Model 2 (adjusted for demographic characteristics including age, gender, marital status, and residence); and Model 3 (additionally adjusted for smoking history, drinking history, BMI, hypertension, and diabetes). The low SES group served as the reference category. Time-varying confounding factors were addressed through baseline covariate measurement (2011), which is methodologically appropriate for evaluating how baseline socioeconomic status predicts future arthritis incidence. While certain covariates (e.g., BMI, hypertension status) may evolve during the 7-year follow-up period, our analytical focus on baseline characteristics aligns with established epidemiological approaches for assessing exposure-outcome relationships and avoids potential bias from reverse causation. The proportional hazards assumption was rigorously assessed using Schoenfeld residuals tests at both individual covariate and global model levels. Individual tests evaluated whether hazard ratios remained temporally constant, with P < 0.05 indicating assumption violation. Global testing assessed overall model appropriateness, with P > 0.05 confirming adequate model specification. Analysis confirmed assumption satisfaction for the primary exposure variable, socioeconomic status (χ² = 0.596, df = 3, P = 0.897), and globally across all model covariates (χ² = 11.112, df = 12, P = 0.519), thereby validating our Cox regression methodology. Kaplan-Meier analysis was performed to estimate the cumulative incidence of arthritis across different SES groups over the follow-up period. The log-rank test was used to compare differences in cumulative arthritis incidence curves between SES groups. Subgroup analyses were performed stratified by gender and age (45–69 years vs. ≥70 years) to evaluate potential effect modification. Interactions between SES and subgroup variables were assessed using likelihood ratio tests comparing nested models with and without interaction terms. The low SES category served as the reference group. Sensitivity analyses were conducted using alternative SES classifications to test the robustness of our findings. Three approaches were employed: binary classification (low/low-middle vs. upper-middle/high SES), tertile classification maintaining low SES as reference, and linear trend analysis treating SES as an ordinal variable. Statistical significance was set at P < 0.05. Statistical analyses were performed using R software (version 4.3.1).
Results
Baseline characteristics of participants in the CHARLS cohort
Of the 4,469 participants included in the analysis (Table 1), 915 (20.5%) were classified as low SES, 2,318 (51.9%) as low-middle SES, 1,201 (26.9%) as upper-middle SES, and 35 (0.8%) as high SES. The age distribution differed significantly across SES groups (P < 0.001), showing a gradient from low (61.44 ± 9.31 years) through low-middle (57.46 ± 8.50 years) and upper-middle (56.13 ± 8.22 years) to high SES groups (57.69 ± 10.65 years). Gender distribution showed significant variation across SES categories (P < 0.001), with male representation increasing progressively from low SES (430, 46.99%) through low-middle (1,135, 48.96%) and upper-middle (652, 54.29%) to high SES groups (26, 74.29%). BMI distribution demonstrated significant differences (P < 0.001), with overweight prevalence increasing across SES groups from low SES (214, 23.39%) to high SES (15, 42.86%). Urban residence showed a marked socioeconomic gradient (P < 0.001), increasing from low SES (197, 21.53%) to high SES (33, 94.29%). Marriage rates exhibited significant differences across SES strata (P < 0.001), with married participants ranging from 756 (82.62%) in the low SES group to 33 (94.29%) in the high SES group. Alcohol consumption patterns varied significantly (P = 0.019), with the highest proportion in the high SES group (23, 65.71%) compared to other groups. Diabetes prevalence showed significant variation (P = 0.030), with rates of 5.68 ± 0.76% in the low SES group compared to 14.29 ± 5.91% in the high SES group. During follow-up, arthritis incidence demonstrated a significant inverse association with SES (P < 0.001). The incidence decreased from 316 (34.54%) in the low SES group to 5 (14.29%) in the high SES group. No significant differences were observed in smoking history (P = 0.138) or hypertension prevalence (P = 0.699) across SES groups.
Table 1.
Baseline characteristics of the study participants in the CHARLS
| Social economic status | P-value | ||||
|---|---|---|---|---|---|
| Low | Low-middle | Upper-middle | High | ||
| Participants (n) | 915 | 2318 | 1201 | 35 | |
| Age, y(mean ± SD) | 61.44 ± 9.31 | 57.46 ± 8.50 | 56.13 ± 8.22 | 57.69 ± 10.65 | < 0.001 |
| Gender (n,%) | < 0.001 | ||||
| Male | 430 (46.99%) | 1135 (48.96%) | 652 (54.29%) | 26 (74.29%) | |
| Female | 485 (53.01%) | 1183 (51.04%) | 549 (45.71%) | 9 (25.71%) | |
| BMI (n,%) | < 0.001 | ||||
| Underweight(< = 18.5) | 84 (9.18%) | 139 (6.00%) | 50 (4.16%) | 0 (0.00%) | |
| Normal(> 18.5, <=24) | 535 (58.47%) | 1259 (54.31%) | 592 (49.29%) | 15 (42.86%) | |
| Overweight(> 24, <=28) | 214 (23.39%) | 674 (29.08%) | 415 (34.55%) | 15 (42.86%) | |
| Obese(> 28) | 82 (8.96%) | 246 (10.61%) | 144 (11.99%) | 5 (14.29%) | |
| Smoking history (n,%) | 0.138 | ||||
| No | 518 (56.61%) | 1321 (56.99%) | 730 (60.78%) | 20 (57.14%) | |
| Yes | 397 (43.39%) | 997 (43.01%) | 471 (39.22%) | 15 (42.86%) | |
| Drinking history (n,%) | 0.019 | ||||
| No | 556 (60.77%) | 1400 (60.40%) | 723 (60.20%) | 12 (34.29%) | |
| Yes | 359 (39.23%) | 918 (39.60%) | 478 (39.80%) | 23 (65.71%) | |
| Residence (n,%) | < 0.001 | ||||
| Urban | 197 (21.53%) | 596 (25.71%) | 651 (54.20%) | 33 (94.29%) | |
| Rural | 718 (78.47%) | 1722 (74.29%) | 550 (45.80%) | 2 (5.71%) | |
| Marital status (n,%) | < 0.001 | ||||
| Married | 756 (82.62%) | 2093 (90.29%) | 1120 (93.26%) | 33 (94.29%) | |
| Single | 159 (17.38%) | 225 (9.71%) | 81 (6.74%) | 2 (5.71%) | |
| Hypertension (n,%) | 0.699 | ||||
| No | 705 (77.05%) | 1814 (78.26%) | 932 (77.60%) | 25 (71.43%) | |
| Yes | 210 (22.95%) | 504 (21.74%) | 269 (22.40%) | 10 (28.57%) | |
| Diabetes (n,%) | 0.030 | ||||
| No | 863 (94.32%) | 2210 (95.34%) | 1129 (94.00%) | 30 (85.71%) | |
| Yes | 52 (5.68%) | 108 (4.66%) | 72 (6.00%) | 5 (14.29%) | |
| Arthritis during follow-up (n,%) | < 0.001 | ||||
| No | 599 (65.46%) | 1597 (68.90%) | 919 (76.52%) | 30 (85.71%) | |
| Yes | 316 (34.54%) | 721 (31.10%) | 282 (23.48%) | 5 (14.29%) | |
Association between social economic status and risk of arthritis: multivariable binary logistic regression analysis
The association between SES and arthritis risk was examined through three progressive models (Table 2). In the unadjusted analysis (Model 1), compared with the low SES group, a significant protective gradient was observed across increasing SES levels, with the strongest protection in the high SES group (OR = 0.32, 95%CI: 0.11–0.75, P = 0.018) and substantial risk reduction in the upper-middle SES group (OR = 0.58, 95%CI: 0.48–0.70, P < 0.001). After adjusting for demographic characteristics (Model 2), the protective association remained robust for the upper-middle SES group (OR = 0.66, 95%CI: 0.54–0.81, P < 0.001), while the high SES group maintained a similar magnitude of protection albeit with marginal significance (OR = 0.41, 95%CI: 0.13–0.97, P = 0.069). Further adjustment for lifestyle factors and comorbidities (Model 3) yielded consistent results (upper-middle: OR = 0.67, 95%CI: 0.55–0.82, P < 0.001; high: OR = 0.41, 95%CI: 0.14-1.00, P = 0.074), suggesting an independent protective effect of higher SES against arthritis development, particularly in the upper-middle SES group.
Table 2.
Association between SES and arthritis risk: results from multivariable logistic regression models
| SES | Participants (n) | Arthritis cases (n) | Non-adjusted | P-value | Adjust I | P-value | Adjust II | P-value |
|---|---|---|---|---|---|---|---|---|
| Low | 915 | 316 | Ref | Ref | Ref | |||
| Low-middle | 2,318 | 721 | 0.86 (0.73–1.01) | 0.060 | 0.90(0.76–1.06) | 0.202 | 0.90(0.76–1.06) | 0.194 |
| Upper-middle | 1,201 | 282 | 0.58 (0.48–0.70) | < 0.001 | 0.66(0.54–0.81) | < 0.001 | 0.67(0.55–0.82) | < 0.001 |
| High | 35 | 5 | 0.32 (0.11–0.75) | 0.018 | 0.41(0.13–0.97) | 0.069 | 0.41(0.14-1.00) | 0.074 |
Model 1: Non-adjusted
Model 2 (Adjust I): Adjusted for demographic characteristics (age, gender, marital status, and residence)
Model 3 (Adjust II): Additionally adjusted for lifestyle factors and comorbidities (smoking history, drinking history, hypertension, diabetes, and body mass index)
Data are presented as odds ratios (ORs) with 95% confidence intervals (CIs). Reference group: participants without low SES
Association between social economic status and risk of arthritis: Cox proportional hazards analysis
SES was significantly associated with the risk of arthritis(Table 3; Fig. 2). Cox proportional hazards analysis revealed that compared with the low SES group, higher SES levels were associated with progressively decreased risks of arthritis, showing a significant gradient effect. In the unadjusted model, the hazard ratios for low-middle, upper-middle, and high SES groups were 0.87 (95% CI: 0.77-1.00, P = 0.047), 0.63 (95% CI: 0.54–0.74, P < 0.001), and 0.36 (95% CI: 0.15–0.86, P = 0.022), respectively. After adjusting for demographic factors (Model 2), this protective association was slightly attenuated but remained. Further adjustment for lifestyle factors and comorbidities (Model 3) showed that the upper-middle SES group maintained a significant protective effect (HR = 0.71, 95% CI: 0.60–0.84, P < 0.001), suggesting that higher SES might be an independent protective factor against arthritis. Although the high SES group demonstrated the strongest protective effect (HR = 0.44, 95% CI: 0.18–1.08), its statistical significance was marginally diminished in the fully adjusted model (P = 0.074), possibly due to the smaller sample size in this category. The proportional hazards assumption of the Cox regression model was assessed using Schoenfeld residuals tests. The assumption was satisfied for the primary exposure variable, social economic status (χ² = 0.596, df = 3, P = 0.897), confirming that the hazard ratios remained constant over time. The global test for all covariates in the model also supported the proportional hazards assumption (χ² = 11.112, df = 12, P = 0.519), validating the appropriateness of the Cox regression approach.
Table 3.
Cox proportional hazards analysis of the association between socioeconomic status and risk of arthritis
| SES levels | Participants (n) | Arthritis cases (n) | Non-adjusted | P-value | Adjust I | P-value | Adjust II | P-value |
|---|---|---|---|---|---|---|---|---|
| Low | 915 | 316 | Ref | Ref | Ref | |||
| Low-middle | 2,318 | 721 | 0.87 (0.77-1.00) | 0.047 | 0.91 (0.80–1.04) | 0.173 | 0.91 (0.79–1.04) | 0.166 |
| Upper-middle | 1,201 | 282 | 0.63 (0.54–0.74) | < 0.001 | 0.71 (0.59–0.84) | < 0.001 | 0.71 (0.60–0.84) | < 0.001 |
| High | 35 | 5 | 0.36 (0.15–0.86) | 0.022 | 0.44 (0.18–1.07) | 0.070 | 0.44 (0.18–1.08) | 0.074 |
| Model diagnostics | PH Assumption Test for SESa | χ² = 0.596, df = 3, P = 0.897 | ||||||
| Global PH Assumption Testb | χ² = 11.112, df = 12, P = 0.519 | |||||||
Data are presented as HR (95% CI) from Cox proportional hazards regression models for arthritis risk
Model 1: Unadjusted model;
Model 2 (Adjust I): Adjusted for age, gender, marital status, and location;
Model 3 (Adjust II): Adjusted for smoking history, drinking history, BMI, hypertension, diabetes, age, gender, marital status, and location;
a Schoenfeld residuals test for social economic status variable;
b Global proportional hazards assumption test for the entire model;
df degrees of freedom; PH Proportional hazards
Fig. 2.
Associations between SES and risk of arthritis across demographic subgroups. Notes: Forest plot showing adjusted HRs and 95% CIs for arthritis risk across different SES levels, stratified by demographic characteristics. Low SES group served as the reference
Social economic status and risk of arthritis: a 7-year cumulative incidence analysis
Kaplan-Meier survival analysis demonstrated significant differences in cumulative arthritis incidence across SES groups during the 7-year follow-up period (log-rank test, p < 0.0001) (Fig. 3). The cumulative incidence was 34.5% in the low SES group, 31.1% in the low-middle SES group, 23.5% in the upper-middle SES group, and 14.3% in the high SES group. Temporal analysis revealed comparable cumulative incidence across groups during the first 2 years, after which between-group differences emerged and progressively widened. Compared with the low SES group, the high SES group showed a 58.6% reduction in cumulative incidence (HR = 0.41, 95% CI: 0.35–0.48, p < 0.0001).
Fig. 3.
Cumulative incidence of arthritis by social economic status during 7-year follow-up. Notes: Kaplan-Meier curves showing cumulative arthritis incidence stratified by SES. X-axis: follow-up time in years; Y-axis: cumulative incidence of arthritis. Log-rank test P < 0.0001
Subgroup analysis of social economic status and arthritis risk
Subgroup analyses were conducted to examine potential effect modification by gender and age on the association between SES and incident arthritis risk (Table 4; Fig. 4). Gender-stratified analyses demonstrated consistent dose-response protective effects across both sexes, with no evidence of effect modification by gender (P for interaction = 0.271). Among male participants, those in the upper-middle SES category exhibited a statistically significant 34% reduction in arthritis risk relative to the low SES reference group (HR = 0.66, 95% CI: 0.52–0.83, P < 0.001). Female participants in the upper-middle SES group demonstrated a comparable magnitude of protection, with a 37% risk reduction (HR = 0.63, 95% CI: 0.50–0.78, P < 0.001). Although the high SES group showed a protective trend in males (HR = 0.56, 95% CI: 0.23–1.36, P = 0.199), hazard ratios were not estimable for females due to zero incident cases in this stratum. Age-stratified analyses revealed more pronounced SES-related protective effects in middle-aged participants (45–69 years) compared with older adults (≥ 70 years), although formal testing showed no significant interaction (P for interaction = 0.371). In the younger age stratum, upper-middle SES conferred a significant 38% reduction in arthritis risk (HR = 0.62, 95% CI: 0.52–0.73, P < 0.001), while high SES demonstrated an even more substantial 76% risk reduction (HR = 0.24, 95% CI: 0.08–0.76, P = 0.015). Conversely, protective effects were markedly attenuated in participants aged ≥ 70 years, with no SES category achieving statistical significance. The absence of significant interaction terms indicates that socioeconomic gradients in arthritis risk remain robust across demographic subgroups, supporting the generalizability of these protective associations.
Table 4.
Subgroup analysis of social economic status on arthritis risk with interaction tests
| Subgroup | Social economic status | Participants (n) | Events (n) | Person-years | Incidence rate* | HR (95% CI) | P value | P for interaction |
|---|---|---|---|---|---|---|---|---|
| By gender | 0.271 | |||||||
| Male | Low SES | 430 | 148 | 2,856 | 51.8 | 1.00 (Reference) | Reference | |
| Low-middle SES | 1,135 | 312 | 7,645 | 40.8 | 0.88 (0.72–1.08) | 0.227 | ||
| Upper-middle SES | 652 | 156 | 4,521 | 34.5 | 0.66 (0.52–0.83) | < 0.001 | ||
| High SES | 26 | 2 | 180 | 11.1 | 0.56 (0.23–1.36) | 0.199 | ||
| Female | Low SES | 485 | 168 | 3,214 | 52.3 | 1.00 (Reference) | Reference | |
| Low-middle SES | 1,183 | 409 | 7,958 | 51.4 | 0.87 (0.73–1.04) | 0.129 | ||
| Upper-middle SES | 549 | 126 | 3,805 | 33.1 | 0.63 (0.50–0.78) | < 0.001 | ||
| High SES | 9 | 0 | 63 | 0.0 | Not estimable† | — | ||
| By age | 0.371 | |||||||
| 45–69 years | Low SES | 658 | 216 | 4,425 | 48.8 | 1.00 (Reference) | Reference | |
| Low-middle SES | 1,846 | 572 | 12,434 | 46.0 | 0.87 (0.75-1.00) | 0.055 | ||
| Upper-middle SES | 1,014 | 238 | 7,033 | 33.8 | 0.62 (0.52–0.73) | < 0.001 | ||
| High SES | 28 | 1 | 194 | 5.2 | 0.24 (0.08–0.76) | 0.015 | ||
| ≥ 70 years | Low SES | 257 | 100 | 1,645 | 60.8 | 1.00 (Reference) | Reference | |
| Low-middle SES | 472 | 149 | 3,169 | 47.0 | 0.91 (0.65–1.27) | 0.569 | ||
| Upper-middle SES | 187 | 44 | 1,293 | 34.0 | 0.77 (0.49–1.23) | 0.277 | ||
| High SES | 7 | 1 | 49 | 20.4 | 1.17 (0.29–4.79) | 0.825 |
*Incidence rate per 1,000 person-years
†Not estimable due to zero arthritis events in the female high SES subgroup
All models were adjusted for age, gender, marital status, residence, smoking history, drinking history, hypertension, diabetes, and body mass index
P for interaction was calculated using likelihood ratio test
HR Hazard ratio; CI Confidence interval; SES Socioeconomic status
Fig. 4.
Subgroup analysis: association between social economic status and arthritis risk
Sensitivity analyses of social economic status and arthritis risk
Sensitivity analyses examining alternative SES classifications confirmed the robustness of our findings (Table 5). Binary classification showed high SES significantly associated with reduced arthritis risk (HR = 0.756, 95% CI: 0.659–0.868, P < 0.001). Tertile classification demonstrated significant protection for high SES (HR = 0.704, 95% CI: 0.594–0.835, P < 0.001), while middle SES showed a non-significant trend (HR = 0.909, 95% CI: 0.794–1.041, P = 0.167). Linear trend analysis confirmed a dose-response relationship (HR = 0.842 per unit increase, 95% CI: 0.775–0.915, P < 0.001). These consistent findings across different SES operationalizations support the robustness of the observed socioeconomic gradient in arthritis risk.
Table 5.
Sensitivity analyses for the association between social economic status and arthritis risk
| SES classification | Comparison | HR (95% CI) | P-value |
|---|---|---|---|
| Binary classification | |||
| High SES vs. Low SES | 0.756 (0.659–0.868) | < 0.001 | |
| Tertile classification | |||
| Middle SES vs. Low SES | 0.909 (0.794–1.041) | 0.167 | |
| High SES vs. Low SES | 0.704 (0.594–0.835) | < 0.001 | |
| Tertile-based linear trend | |||
| Per unit increase in SES | 0.842 (0.775–0.915) | < 0.001 | |
Data are presented as HR (95% CI) from Cox proportional hazards regression models for arthritis risk
Binary classification: Low SES group (Low + Low-middle SES) vs. High SES group (Upper-middle + High SES)
Tertile classification: Low SES vs. Middle SES (Low-middle SES) vs. High SES (Upper-middle + High SES)
Tertile-based linear trend: SES tertile treated as ordinal variable (1 = Low, 2 = Middle, 3 = High SES)
All models adjusted for age, gender, marital status, location, smoking history, drinking history, BMI, hypertension, and diabetes
Discussion
The impact of SES on health outcomes represents a frontier in epidemiological research, with mechanisms underlying its role in chronic disease development requiring further elucidation. This study utilized the CHARLS to establish a large-scale prospective cohort comprising 4,469 arthritis-free middle-aged and elderly participants, conducting 7-year continuous follow-up to investigate the longitudinal association between SES (exposure variable) and new-onset arthritis (outcome variable). Core analytical results demonstrated consistent and significant socioeconomic protective gradients. Multivariable logistic regression analysis showed that in the fully adjusted model, the upper-middle SES group exhibited a 33% risk reduction compared with the low SES group (OR = 0.67, 95% CI: 0.55–0.82, P < 0.001), while the high SES group demonstrated even stronger protection (OR = 0.41, 95% CI: 0.14-1.00, P = 0.074). Cox proportional hazards regression further confirmed that upper-middle SES conferred a 29% reduction in arthritis risk compared with low SES (HR = 0.71, 95% CI: 0.60–0.84, P < 0.001), with the high SES group showing a 56% risk reduction (HR = 0.44, 95% CI: 0.18–1.08, P = 0.074). Kaplan-Meier survival analysis revealed that 7-year cumulative incidence decreased significantly from 34.5% in the low SES group to 14.3% in the high SES group, demonstrating a clear dose-response relationship (log-rank P < 0.001). Subgroup analyses confirmed that these protective effects remained consistent across both males and females (P for interaction = 0.271) and were more pronounced in middle-aged participants (45–69 years). This finding provides important empirical evidence for the applicability of socioeconomic health gradient theory in China’s aging society and establishes a scientific foundation for developing precision arthritis prevention strategies based on population social stratification.
The robustness of our primary findings was further validated through comprehensive sensitivity analyses employing alternative SES classification strategies. Consistent protective effects observed across binary, tertile, and linear trend approaches indicate that the socioeconomic gradient in arthritis risk is not dependent on specific categorization methods, strengthening confidence in our conclusions. The protective effect of SES observed in our study strongly aligns with previous research. Reyes et al.‘s ecological study of 5.47 million Spanish residents demonstrated that the lowest socioeconomic areas had 1.2–1.5 times higher incidence rates of hand, hip, and knee osteoarthritis compared to the highest status areas, revealing clear socioeconomic gradient effects [15]. Two large-scale Korean studies further confirmed this association: Yang et al. tracked over one million individuals for 11 years and found that rheumatoid arthritis incidence remained consistently higher in low SES populations, with this disparity stable throughout 2001-2011 [21]; Lee et al., using national health survey data, confirmed that low education (OR = 1.52), low income (OR = 1.31), and non-managerial positions (OR = 1.28) significantly increased knee osteoarthritis risk [22]. In a cross-sectional survey across six low- and middle-income countries, Brennan-Olsen et al. similarly observed significantly higher arthritis prevalence among socially disadvantaged populations, with this association persisting after age and sex adjustment [23].
Compared to ecological studies, our individual-level longitudinal design enables precise adjustment for key confounders including demographic characteristics, lifestyle factors, and comorbidities, while encompassing a broader arthritis spectrum with stronger causal inference capability. However, some studies reported different findings. Witkam et al.‘s UK Biobank study of 460,000 participants used structural equation modeling and found that SES protection against osteoarthritis was primarily mediated through BMI, with direct effects of SES significantly attenuated after BMI adjustment [24]. This study observed 28,000 incident osteoarthritis cases during follow-up. This discrepancy likely stems from fundamental population differences: UK Biobank predominantly includes relatively healthy middle-aged individuals with higher obesity rates (~ 27%), while our CHARLS longitudinal study encompasses broader elderly populations where malnutrition and underweight conditions are equally common. Therefore, SES protection mechanisms are more multifaceted, extending beyond weight regulation. Izadi et al.‘s multi-ethnic rheumatoid arthritis cohort study found that low SES was associated with functional deterioration, but this study focused on disease progression rather than incident case risk, with fundamental differences in study endpoints potentially explaining the divergent results [25].
SES influences arthritis development through two primary biological pathways. First, socioeconomic disparities in nutritional access and occupational exposure establish contrasting inflammatory environments. Higher SES populations benefit from enhanced access to ω−3 fatty acid-rich foods, which inhibit cyclooxygenase-2 (COX-2) and lipoxygenase (LOX) activity, reducing pro-inflammatory mediators prostaglandin E2 (PGE2) and leukotriene B4 (LTB4) [26, 27]. The ω−3 metabolite resolvin D1 directly activates anti-inflammatory receptors and suppresses interleukin-1β (IL-1β), tumor necrosis factor-α (TNF-α), and interleukin-6 (IL-6) transcription, protecting articular cartilage from inflammatory damage [28]. Conversely, low SES populations face greater occupational mechanical stress, where repetitive loading activates mechanosensitive TRPV4 channels and p38 MAPK signaling pathways in chondrocytes, upregulating matrix metalloproteinase expression and accelerating cartilage degradation [29, 30].
Second, chronic psychosocial stress amplifies joint destruction through hypothalamic-pituitary-adrenal (HPA) axis-mediated inflammatory cascades. Low SES-associated chronic stress activates the corticotropin-releasing hormone (CRH)-adrenocorticotropic hormone (ACTH)-cortisol neuroendocrine axis [31]. Under sustained stress, glucocorticoid resistance leads to persistent NF-κB and activator protein-1 (AP-1) transcription factor activation, upregulating pro-inflammatory cytokine gene expression [32]. The resulting inflammatory mediators orchestrate joint destruction: IL-1β activates NF-κB signaling in synovial fibroblasts to upregulate MMP-1/3/13 expression; TNF-α stimulates macrophage activation and additional pro-inflammatory mediator release; IL-6 levels correlate positively with psychosocial stressors, perpetuating inflammatory cycles [33, 34]. This molecular cascade ultimately culminates in synovial inflammation, cartilage matrix degradation, and bone loss.
These biological pathways operate synergistically rather than independently. Nutritional deficiencies common in low-SES populations impair the body’s natural anti-inflammatory and tissue repair mechanisms, while simultaneously increasing susceptibility to occupational and psychosocial stressors. Chronic stress further disrupts nutritional absorption and metabolic processes, creating a self-perpetuating cycle where socioeconomic disadvantage leads to biological vulnerability, which in turn reduces work capacity and further limits socioeconomic mobility. This interconnected pathway system explains why single-factor interventions may be less effective than comprehensive approaches addressing multiple determinants simultaneously.
This study established the first causal association between SES and arthritis incidence in Chinese middle-aged and elderly populations through 7-year CHARLS cohort tracking, demonstrating a significant 56% risk reduction in high-SES individuals (HR = 0.44) and filling critical evidence gaps in Asian populations. The revealed protective SES gradient effect provides important translational value: integrating SES assessment into risk stratification systems and implementing targeted early screening with multifaceted prevention strategies for low-SES populations. Simultaneously, this study offers evidence-based foundations for policies addressing health inequalities, suggesting that enhanced healthcare accessibility and strengthened occupational protection can effectively promote health equity. Future research should elucidate the biological mechanisms underlying SES influences on arthritis and develop individualized stratified prevention models.
Our findings have important implications for arthritis prevention strategies. Healthcare systems should integrate socioeconomic assessment into routine arthritis risk evaluation, enabling targeted early intervention for high-risk populations [35]. At the policy level, workplace safety regulations and ergonomic standards could reduce occupational arthritis risk among manual laborers, while expanded access to nutritional support programs could address dietary determinants [36]. Community-based interventions combining affordable physical activity programs, stress management resources, and nutritional education may offer cost-effective approaches to reducing socioeconomic disparities in arthritis risk [37]. Future research should evaluate the effectiveness and cost-efficiency of these multi-level intervention strategies through randomized controlled trials to inform evidence-based policy development.
This study has several limitations. First, the determination of new-onset arthritis based on self-reported physician diagnosis may introduce misclassification bias. Participants with lower socioeconomic status might be less likely to seek medical care for arthritis symptoms due to financial constraints or limited healthcare access, potentially leading to underdiagnosis in this population. Conversely, individuals with higher SES may have greater healthcare utilization, resulting in higher diagnostic detection rates. Additionally, some participants might have received diagnoses from non-physician providers or may have had undiagnosed symptomatic arthritis, which could affect the observed SES gradient. However, self-reported physician diagnosis remains the most feasible approach for large-scale longitudinal studies and has been validated in previous epidemiological research. Second, inclusion and exclusion criteria restrict applicability: excluding participants under 45 years limits applicability to younger adults, while excluding those with baseline arthritis restricts results to primary prevention rather than disease progression assessment. The single CHARLS cohort design requires validation through independent cohorts for external validity. The Chinese middle-aged and elderly study population limits applicability to other ethnic groups. Although multiple measurable confounders were adjusted, unmeasurable factors such as genetic susceptibility and early life experiences may still affect result interpretation.
Conclusion
This study established the first causal association between SES and arthritis incidence in Chinese middle-aged and elderly populations using CHARLS cohort data. The findings suggest integrating SES assessment into arthritis risk stratification and implementing targeted prevention strategies for low-SES populations. This study provides evidence-based support for public health policies aimed at reducing health inequalities. Future research should explore the mechanisms underlying SES effects on arthritis and validate intervention efficacy.
Supplementary Information
Acknowledgements
We thank the China Health and Retirement Longitudinal Study (CHARLS) team for providing access to the database and methodological guidance. We sincerely acknowledge all participants who contributed their time and personal information to this study.
Abbreviations
- ACTH
Adrenocorticotropic hormone
- ANOVA
Analysis of variance
- AP
1-Activator protein-1
- BMI
Body mass index
- CHARLS
China Health and Retirement Longitudinal Study
- CI
Confidence interval
- COX
2-Cyclooxygenase-2
- CRH
Corticotropin-releasing hormone
- DALYs
Disability-adjusted life years
- GBD
Global Burden of Disease
- HPA
Hypothalamic-pituitary-adrenal
- HR
Hazard ratio
- IL
1β-Interleukin-1β
- IL
6-Interleukin-6
- IRB
Institutional Review Board
- LOX
Lipoxygenase
- LTB4
Leukotriene B4
- MAPK
Mitogen-activated protein kinase
- MMP
Matrix metalloproteinase
- NF
κB-Nuclear factor kappa B
- OR
Odds ratio
- PGE2
Prostaglandin E2
- SES
Socioeconomic status
- TNF
α-Tumor necrosis factor-α
- TRPV4
Transient receptor potential vanilloid 4
Author contributions
Research conception and methodological framework originated from CL THF with CZY. Statistical evaluation and preliminary writing were undertaken by CL alongside THF. CZY delivered supervisory guidance while refining content quality. Every team member scrutinized outcomes, engaged in textual enhancements, and sanctioned the ultimate manuscript for journal submission.
Funding
This work was supported by the Quanzhou Science and Technology Program (Grant No. 2021N067S & 2025QZNY067). The funding body had no involvement in study design, data collection, analysis, interpretation, or manuscript preparation.
Data availability
The datasets used and analyzed during the current study are available from the China Health and Retirement Longitudinal Study (CHARLS) repository. Access to the data requires registration at http://charls.pku.edu.cn/. Additional data used in this study are included within the article and supplementary materials.
Declarations
Ethics approval and consent to participate
The China Health and Retirement Longitudinal Study (CHARLS) received ethics approval from the Institutional Review Board (IRB) of Peking University (IRB00001052-11015). All participants provided written informed consent before participation. This study utilized publicly available de-identified data from CHARLS and did not require additional ethics approval.
Consent for publication
Not applicable. This study utilized publicly available secondary data from the China Health and Retirement Longitudinal Study (CHARLS), which does not contain any individual personal data requiring specific consent for publication.
Competing interests
The authors declare no competing interests.
Footnotes
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- 1.GBD 2021 Osteoarthritis Collaborators. Global, regional, and National burden of osteoarthritis, 1990–2020 and projections to 2050: a systematic analysis for the global burden of disease study 2021. Lancet Rheumatol. 2023;5:e508–22. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Wang Z, et al. An explainable machine learning-based prediction model for sarcopenia in elderly Chinese people with knee osteoarthritis. Aging Clin Exp Res. 2025;37:67. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Liu S, et al. Global burden of musculoskeletal disorders and attributable factors in 204 countries and territories: a secondary analysis of the global burden of disease 2019 study. BMJ Open. 2022;12:e062183. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Tan MP. Healthcare for older people in lower and middle income countries. Age Ageing. 2022;51:afac016. [DOI] [PubMed] [Google Scholar]
- 5.Hill-Briggs F, et al. Social determinants of health and diabetes: a scientific review. Diabetes Care. 2020;44:258–79. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Navarro-Carrillo G, Alonso-Ferres M, Moya M. Valor-Segura, I. socioeconomic status and psychological Well-Being: revisiting the role of subjective socioeconomic status. Front Psychol. 2020;11:1303. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Wang J, Geng L. Effects of socioeconomic status on physical and psychological health: lifestyle as a mediator. Int J Environ Res Public Health. 2019;16:281. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Braveman P, Gottlieb L. The social determinants of health: it’s time to consider the causes of the causes. Public Health Rep. 2014;129(Suppl 2):19–31. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Barakat C, Konstantinidis TA. Review of the relationship between socioeconomic status change and health. Int J Environ Res Public Health. 2023;20:6249. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Jung M-H, et al. Socioeconomic status and cardiovascular mortality in over 170,000 cancer survivors. Eur Heart J Qual Care Clin Outcomes. 2024:qcae055. 10.1093/ehjqcco/qcae055. [DOI] [PubMed]
- 11.Li S, et al. An umbrella review of socioeconomic status and cancer. Nat Commun. 2024;15:9993. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Meulman I, Uiters E, Cloin M, Polder J, Stadhouders N. Systematic review of individual-level, community-level, and healthcare system-level factors contributing to socioeconomic differences in healthcare utilisation in OECD countries with universal health coverage. BMJ Open. 2024;14:e080559. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Dong P, et al. The relationship among socioeconomic status, social support and frailty: is there a gender difference? Aging Clin Exp Res. 2025;37:111. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Adas M, et al. What role do socioeconomic and clinical factors play in disease activity states in rheumatoid arthritis? Data from a large UK early inflammatory arthritis audit. RMD Open. 2024;10:e004180. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Reyes C, et al. Socio-economic status and the risk of developing hand, hip or knee osteoarthritis: a region-wide ecological study. Osteoarthritis Cartilage. 2015;23:1323–9. [DOI] [PubMed] [Google Scholar]
- 16.Reyes AM, Katz JN. Racial/Ethnic and socioeconomic disparities in osteoarthritis management. Rheum Dis Clin North Am. 2021;47:21–40. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Zhao Y, Hu Y, Smith JP, Strauss J, Yang G. Cohort profile: the China health and retirement longitudinal study (CHARLS). Int J Epidemiol. 2014;43:61–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Zhou Y, et al. Associations between socioeconomic inequalities and progression to psychological and cognitive multimorbidities after onset of a physical condition: a multicohort study. EClinicalMedicine. 2024;74:102739. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Wang D, et al. Association between socioeconomic status and health behaviour change before and after non-communicable disease diagnoses: a multicohort study. Lancet Public Health. 2022;7:e670–82. [DOI] [PubMed] [Google Scholar]
- 20.Zhou B-F. Effect of body mass index on all-cause mortality and incidence of cardiovascular diseases–report for meta-analysis of prospective studies open optimal cut-off points of body mass index in Chinese adults. Biomed Environ Sci. 2002;15:245–52. [PubMed] [Google Scholar]
- 21.Yang D-H, Huang J-Y, Chiou J-Y, Wei J. C.-C. Analysis of socioeconomic status in the patients with rheumatoid arthritis. Int J Environ Res Public Health. 2018;15:1194. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Lee JY, Han K, Park YG, Park S-H. Effects of education, income, and occupation on prevalence and symptoms of knee osteoarthritis. Sci Rep. 2021;11:13983. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Brennan-Olsen SL, et al. Prevalence of arthritis according to age, sex and socioeconomic status in six low and middle income countries: analysis of data from the world health organization study on global ageing and adult health (SAGE) wave 1. BMC Musculoskelet Disord. 2017;18:271. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Witkam R, et al. Does body mass index mediate the relationship between socioeconomic position and incident osteoarthritis? Semin Arthritis Rheum. 2022;56:152063. [DOI] [PubMed] [Google Scholar]
- 25.Izadi Z, et al. Socioeconomic disparities in functional status in a national sample of patients with rheumatoid arthritis. JAMA Netw Open. 2021;4:e2119400. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Calder PC. Omega-3 fatty acids and inflammatory processes. Nutrients. 2010;2:355–74. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Zainal Z, et al. Relative efficacies of omega-3 polyunsaturated fatty acids in reducing expression of key proteins in a model system for studying osteoarthritis. Osteoarthritis Cartilage. 2009;17:896–905. [DOI] [PubMed] [Google Scholar]
- 28.Norling LV, et al. Proresolving and cartilage-protective actions of resolvin D1 in inflammatory arthritis. JCI Insight. 2016;1:e85922. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Zhang K, et al. Mechanosensory and mechanotransductive processes mediated by ion channels in articular chondrocytes: potential therapeutic targets for osteoarthritis. Channels (Austin). 2021;15:339–59. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Li Z, et al. P38MAPK signaling pathway in osteoarthritis: pathological and therapeutic aspects. J Inflamm Res. 2022;15:723–34. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Herman JP, et al. Regulation of the Hypothalamic-Pituitary-Adrenocortical stress response. Compr Physiol. 2016;6:603–21. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Miller GE, Chen E, Parker KJ. Psychological stress in childhood and susceptibility to the chronic diseases of aging: moving toward a model of behavioral and biological mechanisms. Psychol Bull. 2011;137:959–97. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Choi M-C, Jo J, Park J, Kang HK, Park Y. NF-κB signaling pathways in osteoarthritic cartilage destruction. Cells. 2019;8:734. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Zhang H, et al. Synovial macrophage M1 polarisation exacerbates experimental osteoarthritis partially through R-spondin-2. Ann Rheum Dis. 2018;77:1524–34. [DOI] [PubMed] [Google Scholar]
- 35.Rethorn ZD, et al. Association of burden and prevalence of arthritis with disparities in social risk factors, findings from 17 US states. Prev Chronic Dis. 2022;19:E08. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Hoe VC, Urquhart DM, Kelsall HL, Zamri EN, Sim MR. Ergonomic interventions for preventing work-related musculoskeletal disorders of the upper limb and neck among office workers. Cochrane Database Syst Rev. 2018;10:CD008570. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Messier SP, et al. Weight-loss and exercise for communities with arthritis in North Carolina (we-can): design and rationale of a pragmatic, assessor-blinded, randomized controlled trial. BMC Musculoskelet Disord. 2017;18:91. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The datasets used and analyzed during the current study are available from the China Health and Retirement Longitudinal Study (CHARLS) repository. Access to the data requires registration at http://charls.pku.edu.cn/. Additional data used in this study are included within the article and supplementary materials.