Association between digestive diseases and symptomatic knee osteoarthritis in Chinese participants: a prospective cohort study Authors: Shile Chen, Youdi Liu, Daocheng Zhu, Wei Xu, Mingxuan Zheng, Aiwu Liu, Bing Wang, Suipeng Chen, Lin Jiao Published: 2026 ABSTRACT AbstractBackgroundDigestive diseases and symptomatic knee osteoarthritis (OA) frequently occur in older adults, yet prospective evidence clarifying their temporal relationship remains limited, particularly in Asian populations. This study examined the relationship between digestive diseases and the risk of incident symptomatic knee OA in a nationally representative cohort of middle‑aged and older Chinese adults.MethodsData were obtained from the China Health and Retirement Longitudinal Study (CHARLS) from 2011 to 2018. A total of 6,281 adults without symptomatic knee OA at baseline were included. Digestive diseases were defined as exposure, and incident symptomatic knee osteoarthritis was the outcome of interest. Kaplan-Meier survival curves with log‑rank tests were used to compare cumulative incidence, and multivariable Cox proportional hazards models were applied to estimate hazard ratios. Sensitivity analyses included propensity score matching (PSM), prespecified subgroup analyses presented as forest plots, and E-value calculations.ResultsA total of 6,281 participants were included in the analytical cohort (mean age, 58.6 ± 8.8 years; 53.5% women), with a median follow‑up of 84 months. Over the observation period, participants reporting digestive diseases experienced a markedly higher cumulative incidence of symptomatic knee OA compared with those without digestive conditions (31.4% versus 20.9%; log‑rank p < 0.01). After adjustment for potential confounders, multivariable Cox models indicated a higher risk of symptomatic knee OA among participants with digestive diseases (adjusted HR = 1.40; 95% CI: 1.25–1.57; p < 0.001). Similar effect estimates were obtained in the PSM (HR = 1.42; 95% CI: 1.23–1.64; p < 0.001). Subgroup analyses showed similar associations, with no statistically significant interactions detected. E‑value was 2.17.ConclusionDigestive diseases are linked to a higher likelihood of symptomatic knee OA in this prospective cohort, supporting a temporal relationship between digestive disorders and the development of symptomatic knee OA.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12889-026-27690-w.Keywords: Digestive diseases, Symptomatic knee osteoarthritis, Prospective cohort study, Propensity score matching, CHARLS Introduction Knee osteoarthritis (OA) is one of the most common chronic musculoskeletal disorders and a leading cause of pain, functional limitation, and disability among middle‑aged and older adults worldwide. Although osteoarthritis affects multiple joint sites, knee involvement represents the most prevalent and disabling clinical phenotype and contributes the largest share of OA‑related health loss [1–3]. Increasing evidence indicates that knee OA is a multifactorial, chronic disabling condition involving pathological changes across multiple joint structures, rather than a disease confined to articular cartilage alone. Structural and functional alterations associated with knee OA extend beyond cartilage degeneration to include subchondral bone remodeling, synovial inflammation, ligament and joint capsule abnormalities, and periarticular muscle dysfunction, which together contribute to pain, stiffness, and progressive loss of mobility [2, 4–6]. Meanwhile, knee OA represents a major and growing public health burden in aging populations, yet its systemic determinants remain incompletely understood. The global burden of knee OA is substantial and continues to increase in parallel with population ageing, rising obesity prevalence, and prolonged life expectancy. Recent global estimates indicate that the prevalence of knee OA reaches approximately 22.9% among adults aged 40 years and older, making it one of the most prevalent musculoskeletal conditions worldwide [7]. In China knee OA represents a particularly important and growing public health challenge. Based on nationally representative data from the China Health and Retirement Longitudinal Study (CHARLS), approximately 8.8% of Chinese adults aged 45 years and older were estimated to have symptomatic knee OA, accounting for more than 4 million lived with disability (YLDs) nationwide [8, 9]. Recent reviews indicate that knee OA has become an increasingly important public health concern in China as the population ages, with women, older individuals, and rural residents bearing a greater share of the disease burden. Knee OA is associated with substantial impairments in mobility, activities of daily living, and quality of life, resulting in considerable health‑care utilization and socioeconomic burden [10, 11]. Emerging evidence suggests a potential link between gastrointestinal health and joint degeneration, often conceptualized as the “gut–joint axis”. Digestive diseases constitute another major category of chronic conditions contributing substantially to global morbidity and healthcare utilization. According to the Global Burden of Diseases, Injuries, and Risk Factors Study (GBD) 2019, digestive diseases accounted for approximately 18.2% of all incident cases and 38.4% of all prevalent cases globally, underscoring their pervasive impact on population health. In 2019 alone, there were an estimated 7.3 billion new cases and 2.9 billion prevalent cases of digestive diseases, resulting in 8 million deaths and 277 million disability-adjusted life-years (DALYs) [12]. These conditions affect individuals across all age groups and geographic regions, though their distribution and underlying causes vary significantly with socioeconomic development. In China, digestive diseases are highly prevalent among middle‑aged and older adults and constitute one of the most common chronic conditions contributing to long‑term morbidity and health‑care utilization [13]. Importantly, digestive diseases are increasingly recognized as systemic disorders rather than isolated gastrointestinal conditions. Recent epidemiological studies have shown that digestive diseases are associated with adverse outcomes such as frailty, sarcopenia, and mental health disorders, reflecting accelerated functional decline and multisystem impairment [14–16]. Notably, several of these systemic disturbances—particularly chronic inflammation, metabolic abnormalities, and impaired physical function—are also key contributors to the pathogenesis of knee OA, providing a biologically plausible basis for a potential association between digestive diseases and knee OA [17, 18]. However, large-scale prospective cohort evidence, particularly longitudinal data systematically evaluating the targeted association in Asian and Chinese populations, remains scarce. Notably, a prior CHARLS-based longitudinal study [11] reported digestive diseases as one of multiple factors associated with symptomatic knee OA outcomes. However, digestive diseases were not treated as the primary exposure, and the association was not presented in a focused time-to-event, robustness-oriented framework. Existing studies have often focused on small clinical samples or individual gastrointestinal conditions and have primarily relied on cross‑sectional designs, with inadequate adjustment for major confounders such as age, obesity, physical activity, comorbid chronic diseases, and socioeconomic status. Therefore, this study aims to examine the association between digestive diseases and the risk of incident symptomatic knee OA in a nationally representative cohort of Chinese adults aged 45 years and older, using comprehensive adjustment for key sociodemographic, lifestyle, laboratory biomarkers, and health‑related covariates to provide more robust epidemiological evidence on the relationship between digestive diseases and symptomatic knee OA and to improve understanding of the systemic determinants of symptomatic knee OA. Materials and methods Data source and study population This prospective cohort analysis utilized data derived from CHARLS, with survey waves conducted between 2011 and 2018. CHARLS is a nationally representative longitudinal survey of Chinese adults aged 45 years and older, designed to collect detailed information on sociodemographic characteristics, health conditions, lifestyle factors, and laboratory measurements. The sampling framework employed a multistage, stratified probability‑based design, allowing for adequate representation of the non‑institutionalized middle‑aged and older population across China. Data collection was carried out through household‑based, face‑to‑face computer‑assisted interviews conducted by trained field staff affiliated with Peking University. This study adhered to the established CHARLS procedures and ethical requirements. Data from CHARLS can be accessed through the official project website (http://charls.pku.edu.cn). For this specific analysis, we included participants aged 45 years or older who were free of symptomatic knee OA at the 2011 baseline and had complete data on digestive disease status and key covariates. This resulted in an analytical cohort of 6,281 participants. Symptomatic knee OA status was assessed at each follow-up wave (2013, 2015, and 2018). Participants were followed from the date of the baseline interview until the first report of symptomatic knee OA, death, loss to follow-up, or the end of follow-up, whichever occurred first. Person-years were calculated accordingly. To further control for confounding, a propensity score-matched sub-cohort was constructed using 1:1 nearest-neighbor matching, yielding 1,354 participants with digestive diseases matched to 1,354 participants without, for a total of 2,708 matched individuals. Assessment of symptomatic knee OA Information on symptomatic knee OA was obtained from self-reported questionnaire data. Symptomatic knee OA was defined using validated criteria that required all the following conditions: (1) a physician diagnosis of arthritis or rheumatism; (2) frequent bodily pain; and (3) self-reported knee pain when pain was present [11, 19]. Participants with a diagnosis but without knee pain were not considered cases at baseline or during follow-up. Assessment of digestive diseases Information on digestive diseases was obtained via self‑reported data. During the interview, participants were queried about a prior physician diagnosis of stomach or other digestive conditions, excluding tumors or cancer. Individuals providing an affirmative response were considered to have digestive diseases [15]. Assessment of covariates Information on covariates was extracted from the CHARLS dataset. To reduce the potential influence of confounding, variables associated with the risk of digestive diseases and symptomatic knee OA were identified and included for adjustment according to existing epidemiological evidence, clinical considerations, and the availability of relevant data within CHARLS [11, 15, 19, 20]. Specifically, sociodemographic characteristics, lifestyle behaviors, metabolic markers, and chronic conditions were identified as key determinants of symptomatic knee OA risk and were included for adjustment to reduce potential confounding. Sociodemographic characteristics included age (categorized as 45–59 years or ≥ 60 years), sex (binary classification: male or female), and marital status (binary operationalization: married [spouse present or temporarily separated] vs. single [separated, divorced, widowed, never married], reflecting social support dynamics. Older age and female sex are well-established, influential risk factors for knee OA across populations [21, 22] while marital status was included to account for social support, which may influence health-seeking behaviors and disease progression. Lifestyle factors comprised body mass index (analyzed continuously as mean ± SD with supplementary categorization using Chinese-specific thresholds: <28 kg/m² vs≥28 kg/m²) [23, 24], drinking status (classified as “Drink > 1/month”, “Drink < 1/month”, or “Never”), smoking status (binary: yes or no), and physical activity quantified by metabolic equivalent tasks (METs). Higher BMI is a consistently strong risk factor for knee OA [21, 22]. Physical activity shows a dose-dependent relationship, with moderate activity appearing protective, whereas excessive mechanical loading may increase knee OA risk [25]. Alcohol consumption has been linked to a higher likelihood of developing knee OA, while smoking has not shown a protective effect against knee OA [26, 27]. Laboratory biomarkers included serum creatinine, HDL cholesterol, hemoglobin, and fasting glucose, all analyzed as continuous variables using standardized CHARLS protocols. These markers reflect underlying systemic states such as muscle mass, lipid metabolism, anemia, and glycemic control, which have been associated with knee OA risk or symptom severity [28–31]. Health status indicators encompassed self-rated health (five-level ordinal scale) and emotional, nervous, or psychiatric problems (binary classification: yes or no), and physician-diagnosed chronic conditions, including heart attack, hypertension, and asthma (binary: yes or no). Psychological distress is common in knee OA and is closely related to pain amplification and functional limitation [32]. Hypertension, cardiovascular diseases, and asthma are more prevalent among individuals with knee OA [33–36], all of which may confound the association between digestive diseases and knee OA. Statistical analyses We employed a complete data analysis by restricting the cohort to participants with no missing values on the exposure, outcome, or covariates. Baseline characteristics of the study population were described using summary statistics and compared between participants with and without digestive diseases. In addition, baseline characteristics were compared between participants included in the analytical cohort and those excluded due to missing covariate data. Continuous variables were summarized as mean ± standard deviation or as median with interquartile range, as appropriate, while categorical variables were reported as frequencies and proportions. The relationship between digestive diseases and the risk of incident symptomatic knee OA was primarily examined using multivariable Cox proportional hazards models. Adjusted hazard ratios (HRs) with 95% confidence intervals (CIs) were calculated after controlling for potential confounding factors. Meanwhile, Kaplan-Meier survival curves were constructed to compare the cumulative incidence of symptomatic knee OA between participants with and without digestive diseases, and differences between groups were assessed using the log-rank test. To assess the robustness of the primary findings, we performed several sensitivity analyses. As a sensitivity analysis, propensity score matching (PSM) was conducted to reduce baseline imbalance and enhance comparability between groups. Propensity scores were estimated using multivariable logistic regression based on prespecified demographic and clinical covariates. Participants were matched 1:1 using nearest-neighbor matching without replacement, applying a caliper of 0.2 standard deviations of the logit of the propensity score. This resulted in a matched cohort of 1,354 participants with digestive diseases and 1,354 participants without. Covariate balance was assessed using standardized mean differences (SMDs), with SMD < 0.1 considered adequate. Changes in balance before and after matching were visualized using a Love plot. In addition, subgroup analyses were performed as sensitivity analyses, and interaction terms were tested within the Cox regression models. We also excluded participants with symptomatic knee OA onset within the first 2 years of follow-up to reduce potential reverse causation, applied the Fine-Gray subdistribution hazard model to address death as a competing risk, and performed E-value analysis to evaluate the robustness of our findings against unmeasured confounding. All statistical analyses were conducted with R software (version 4.2.2; R Foundation for Statistical Computing) and the Free Statistics Analysis Platform (version 2.4.0; Beijing, China). All statistical tests were two-sided, and a p value < 0.05 was considered statistically significant. Results Baseline characteristics of participants As shown in Fig. 1; Table 1, a total of 11,136 participants aged 45 years or older without symptomatic knee OA at baseline in 2011 were initially assessed for eligibility. After excluding 304 participants with missing information on digestive diseases status and 4,551 participants with incomplete covariate data, 6,281 participants were included in the analytical cohort. Baseline characteristics of included and excluded participants are shown in Supplementary Table 1. The mean age of the cohort was 58.6 ± 8.8 years, and 53.5% were female. In the unmatched cohort, participants with digestive diseases differed significantly from those without digestive diseases across several baseline characteristics, including age, sex, BMI, marital status, drinking status, physical activity level, and self-rated health status. Significant differences were also observed in the prevalence of emotional problems and comorbid conditions, including heart attack, hypertension, and asthma. Additionally, multiple laboratory parameters, including creatinine, high-density lipoprotein cholesterol, glucose, and hemoglobin levels, differed significantly between the two groups (all p < 0.05). Flowchart of participants selection Table 1Baseline characteristics of study participants before and after propensity score matching by digestive disease statusVariablesPatient characteristics (raw data)After PSMTotalWithout digestive diseasesWith digestivediseasesP-valueTotalWithout digestive diseasesWith digestivediseasesP-valueN = 6281N = 4917N = 1364N = 2708N = 1354N = 1354Age, years,58.6 ± 8.858.7 ± 8.958.2 ± 8.70.02958.2 ± 8.758.2 ± 8.758.2 ± 8.70.89Sex, n (%)0.0120.439 Male2923 (46.5)2329 (47.4)594 (43.5)1204 (44.5)612 (45.2)592 (43.7) Female3358 (53.5)2588 (52.6)770 (56.5)1504 (55.5)742 (54.8)762 (56.3)Marital status, n (%)0.0020.458 Married678 (10.8)562 (11.4)116 (8.5)241 (8.9)126 (9.3)115 (8.5) Single5603 (89.2)4355 (88.6)1248 (91.5)2467 (91.1)1228 (90.7)1239 (91.5) BMI, kg/m2,23.6 ± 3.923.7 ± 4.023.0 ± 3.6< 0.00123.0 ± 3.623.0 ± 3.723.0 ± 3.60.779Drinking status, n (%)< 0.0010.865 Drink more than once a month1574 (25.1)1287 (26.2)287 (21)578 (21.3)292 (21.6)286 (21.1) Drink but less than once a month501 (8.0)384 (7.8)117 (8.6)238 (8.8)122 (9)116 (8.6) None of these4206 (67.0)3246 (66)960 (70.4)1892 (69.9)940 (69.4)952 (70.3)Smoking, n (%)0.0520.605 Yes2422 (38.6)1927 (39.2)495 (36.3)1001 (37.0)507 (37.4)494 (36.5) No3859 (61.4)2990 (60.8)869 (63.7)1707 (63.0)847 (62.6)860 (63.5) MET0.0 (0.0, 82.2)0.0 (0.0, 79.1)0.0 (0.0, 101.3)0.0420.0 (0.0, 102.2)0.0 (0.0, 102.2)0.0 (0.0, 102.0)0.872Health status, n (%)< 0.0010.91 Excellent202 (3.2)190 (3.9)12 (0.9)24 (0.9)12 (0.9)12 (0.9) Very good769 (12.2)687 (14)82 (6)164 (6.1)82 (6.1)82 (6.1) Good2131 (33.9)1732 (35.2)399 (29.3)784 (29.0)385 (28.4)399 (29.5) Fair2289 (36.4)1715 (34.9)574 (42.1)1173 (43.3)599 (44.2)574 (42.4) Poor890 (14.2)593 (12.1)297 (21.8)563 (20.8)276 (20.4)287 (21.2) Creatinine, mg/dL0.8 ± 0.20.8 ± 0.20.8 ± 0.20.0420.8 ± 0.20.8 ± 0.20.8 ± 0.20.577 HDL Cholesterol, mg/dL51.2 ± 15.350.9 ± 15.252.4 ± 15.4< 0.00152.3 ± 15.352.2 ± 15.352.4 ± 15.40.747 Glucose, mg/dL109.8 ± 35.0110.5 ± 36.8107.3 ± 27.30.003107.4 ± 29.9107.6 ± 32.2107.3 ± 27.30.782 Cholesterol, mg/dL194.3 ± 38.5194.5 ± 38.6193.6 ± 38.20.423193.4 ± 38.0193.3 ± 37.8193.6 ± 38.20.868 Hemoglobin, g/dL14.4 ± 2.214.5 ± 2.314.3 ± 2.10.00214.3 ± 2.214.3 ± 2.214.3 ± 2.10.938Emotional problems nervous, or psychiatric problems, n (%)0.0020.659 No6204 (98.8)4868 (99)1336 (97.9)2661 (98.3)1329 (98.2)1332 (98.4) Yes77 (1.2)49 (1)28 (2.1)47 (1.7)25 (1.8)22 (1.6)Heart attack, n (%)< 0.0010.913 No5648 (89.9)4486 (91.2)1162 (85.2)2318 (85.6)1158 (85.5)1160 (85.7) Yes633 (10.1)431 (8.8)202 (14.8)390 (14.4)196 (14.5)194 (14.3)Hypertension, n (%)0.0250.813 No4831 (76.9)3751 (76.3)1080 (79.2)2141 (79.1)1068 (78.9)1073 (79.2) Yes1450 (23.1)1166 (23.7)284 (20.8)567 (20.9)286 (21.1)281 (20.8)Asthma, n (%)< 0.0010.642 No6083 (96.8)4782 (97.3)1301 (95.4)2587 (95.5)1291 (95.3)1296 (95.7) Yes198 (3.2)135 (2.7)63 (4.6)121 (4.5)63 (4.7)58 (4.3)All normally distributed and skewed continuous variables are presented as medians with interquartile ranges. Categorical variables are expressed as frequencies and percentages (%). P-values: For baseline characteristic analysis, the statistical differences among the quartiles of radiation were tested using one-way analysis of variance for continuous variables and chi-squared tests for categorical variables. Statistical significance was set at p < 0.05SD Standard deviation, IQR Interquartile range, PSM Propensity score–matched, BMI Body mass index, MET Metabolic Equivalent of Task, HDL High-density lipoprotein Baseline characteristics of study participants before and after propensity score matching by digestive disease status All normally distributed and skewed continuous variables are presented as medians with interquartile ranges. Categorical variables are expressed as frequencies and percentages (%). P-values: For baseline characteristic analysis, the statistical differences among the quartiles of radiation were tested using one-way analysis of variance for continuous variables and chi-squared tests for categorical variables. Statistical significance was set at p < 0.05 SD Standard deviation, IQR Interquartile range, PSM Propensity score–matched, BMI Body mass index, MET Metabolic Equivalent of Task, HDL High-density lipoprotein Kaplan-Meier curves analyses As shown in Supplementary Fig. 2A, the median follow-up duration was 84 months. Kaplan–Meier estimates indicated a markedly greater cumulative incidence of symptomatic knee OA among participants with digestive diseases compared with those without digestive diseases (31.4% [428/1,364] vs. 20.9% [1,029/4,917]; log-rank p < 0.001). Multivariable Cox regression analyses As presented in Table 2, a higher incidence of symptomatic knee OA was observed among participants with digestive diseases in the unadjusted analysis (HR = 1.59; 95% CI: 1.42–1.78; p < 0.001). This elevated risk persisted after sequential adjustment for sociodemographic characteristics, lifestyle behaviors, and metabolic factors. In the partially adjusted models, the association remained stable, with hazard ratios of 1.58 (95% CI: 1.41–1.77) in model 1 and 1.57 (95% CI: 1.40–1.76) in model 2 (both p < 0.001). Following additional control for overall health status, emotional conditions, and chronic diseases, digestive diseases continued to show an independent association with incident symptomatic knee OA in the fully adjusted model (HR = 1.40; 95% CI: 1.25–1.57; p < 0.001). Table 2Multivariable Cox regression analysesModelWith digestive diseasesWithout digestive diseasesHR95%CIP-valueTotal, NSymptomatic knee OAevent, n%Total, NSymptomatic knee OAevent, n%Unadjusted model1364428 (31.4%)49171029 (20.9%)1.591.42,1.78<0.001Adjusted model 11.581.41, 1.77<0.001Adjusted model 21.571.40, 1.76<0.001Adjusted model 31.401.25, 1.57<0.001Unadjusted model: no other covariates were adjustedAdjusted model1: we adjusted age, sex, marital status, BMI, alcohol, smoke and METAdjusted model2: we adjusted age, sex, marital status, BMI, alcohol, smoke, MET, creatinine, HDL cholesterol, cholesterol, hemoglobin and glucoseAdjusted model3: we adjusted age, sex, marital status, BMI, alcohol assumption, smoke, MET, creatinine, HDL cholesterol, cholesterol, hemoglobin and glucose, health status, emotional, nervous, or psychiatric problems, heart attack, hypertension and asthma Multivariable Cox regression analyses Unadjusted model: no other covariates were adjusted Adjusted model1: we adjusted age, sex, marital status, BMI, alcohol, smoke and MET Adjusted model2: we adjusted age, sex, marital status, BMI, alcohol, smoke, MET, creatinine, HDL cholesterol, cholesterol, hemoglobin and glucose Adjusted model3: we adjusted age, sex, marital status, BMI, alcohol assumption, smoke, MET, creatinine, HDL cholesterol, cholesterol, hemoglobin and glucose, health status, emotional, nervous, or psychiatric problems, heart attack, hypertension and asthma Sensitivity analyses As a sensitivity analysis, 1:1 propensity score matching was performed, resulting in a matched cohort of 2,708 participants (1,354 with digestive diseases and 1,354 without). After matching, baseline covariates showed no meaningful differences between groups, as presented in Table 1 (all p > 0.05), and all covariates achieved SMDs < 0.1 as shown in Supplementary Fig. 1, suggesting satisfactory balance for subsequent analyses. Kaplan‑Meier analysis in the matched cohort also showed a higher cumulative incidence of symptomatic knee OA among participants with digestive diseases compared with those without (31.5% [426/1,354] vs. 23.3% [315/1,354]; log‑rank p < 0.001), as shown in Supplementary Fig. 2B. To further assess robustness, we examined the association using multiple propensity score‑based methods. As summarized in Table 3, adjusting for the propensity score in the regression model yielded comparable results (HR = 1.41; 95% CI: 1.26–1.58; p < 0.001). In the propensity score–matched cohort comprising 2,708 participants (1,354 per group), digestive diseases continued to be independently associated with incident symptomatic knee OA (HR = 1.42; 95% CI: 1.23–1.64; p < 0.001). Consistent findings were also observed using alternative weighting techniques, including inverse probability weighting, standardized mortality ratio weighting, pairwise algorithmic weighting, and overlap weighting, with effect estimates ranging between 1.36 and 1.41 (all p < 0.001), supporting the robustness of the association. Table 3Associations between digestive diseases and symptomatic knee OA in the propensity-score analysesAnalysisHR (95% CI)P-valueAdjusted for propensity scorea1.41 (1.26 ~ 1.58)< 0.001With matchedb1.42 (1.23 ~ 1.64)< 0.001With inverse probability weightingc1.36 (1.21 ~ 1.53)< 0.001With SMRWd1.40 (1.26 ~ 1.57)< 0.001With PAe1.41 (1.22 ~ 1.63)< 0.001With Owf1.40 (1.18 ~ 1.66)< 0.001Notes: a Shown is the hazard ratio from a multivariable Cox proportional-hazards model with the same strata and covariates, with additional adjustment for the propensity score. b Shown is the hazard ratio from a multivariable Cox proportional-hazards model with the same strata and covariates matching according to the propensity score. The analysis included 2,708 patients (1,354 participants with digestive disease and 1,354 participants without digestive disease)c Shown is the primary analysis with a hazard ratio from the multivariable Cox proportional hazards model with the same strata and covariates with inverse probability weighting according to the propensity scored Shown is the primary analysis with a hazard ratio from the multivariable Cox proportional hazards model with the same strata and covariates with the standardized mortality ratio weighting according to the propensity scoree Shown is the primary analysis with a hazard ratio from the multivariable Cox proportional-hazards model with the same strata and covariates with pairwise algorithmic according to the propensity scoref Shown is the primary analysis with a hazard ratio from the multivariable Cox proportional hazards model with the same strata and covariates with overlap weight according to the propensity score Associations between digestive diseases and symptomatic knee OA in the propensity-score analyses Notes: a Shown is the hazard ratio from a multivariable Cox proportional-hazards model with the same strata and covariates, with additional adjustment for the propensity score. b Shown is the hazard ratio from a multivariable Cox proportional-hazards model with the same strata and covariates matching according to the propensity score. The analysis included 2,708 patients (1,354 participants with digestive disease and 1,354 participants without digestive disease) c Shown is the primary analysis with a hazard ratio from the multivariable Cox proportional hazards model with the same strata and covariates with inverse probability weighting according to the propensity score d Shown is the primary analysis with a hazard ratio from the multivariable Cox proportional hazards model with the same strata and covariates with the standardized mortality ratio weighting according to the propensity score e Shown is the primary analysis with a hazard ratio from the multivariable Cox proportional-hazards model with the same strata and covariates with pairwise algorithmic according to the propensity score f Shown is the primary analysis with a hazard ratio from the multivariable Cox proportional hazards model with the same strata and covariates with overlap weight according to the propensity score As shown in Fig. 2, subgroup analyses demonstrated that digestive diseases were consistently linked to a higher likelihood of symptomatic knee OA across all examined subgroups, both in the crude analysis and after propensity score matching. The magnitude and direction of the associations were generally consistent across sex, age, marital status, BMI (< 28 kg/m² vs. ≥ 28 kg/m²), history of heart attack, and hypertension. No significant interactions were observed in any subgroup analyses. Results were robust after excluding participants who developed symptomatic knee OA within the first 2 years of follow-up and in Fine-Gray models treating death as a competing event (Supplementary Tables 2–3; Supplementary Fig. 3). As shown in Supplementary Fig. 4, E‑value assessment suggested that only an unmeasured confounder with relatively strong associations with both digestive diseases and symptomatic knee OA (risk ratio = 2.17) could fully explain the observed relationship. Association between digestive diseases and symptomatic knee OA across subgroups before and after PSM. This forest plot displays hazard ratios (HRs) with 95% confidence intervals (CIs) for incident symptomatic knee osteoarthritis risk stratified by demographic and clinical subgroups; left panel shows crude associations from unadjusted analyses (total cohort n = 6,281); right panel presents propensity score-matched (PSM) associations after 1:1 matching (caliper = 0.02; matched cohort n = 2,708 with 1,354 per group); square markers indicate point estimates, horizontal lines denote 95% CIs, and vertical dashed line marks null value (HR = 1); all analyses adjusted for 17 covariates (age, sex, marital status, BMI, drinking status, smoking, MET creatinine, HDL cholesterol, total cholesterol, hemoglobin, glucose, emotional problems nervous, or psychiatric problems, heart attack, hypertension and asthma) except when serving as stratification variables; subgroups showed consistent associations (all p-interaction > 0.05) Discussion Using data from a large, nationally representative prospective cohort of middle‑aged and older adults in China, we found that the presence of digestive diseases was linked to an elevated probability of symptomatic knee OA development over time. During a median follow‑up period of 84 months, individuals reporting digestive disorders experienced a greater cumulative incidence of symptomatic knee OA than their counterparts without such conditions. This association remained robust after extensive adjustment for demographic characteristics, lifestyle factors, metabolic indicators, health status, emotional problems, and physician-diagnosed chronic conditions, as well as across multiple propensity score–based approaches. Previous studies have increasingly addressed potential associations between gastrointestinal disorders and arthritis‑associated outcomes. Using cross‑sectional data from NHANES 2011–2018, Lu and Ding [18] reported that participants with gastrointestinal diseases had a significantly higher odds of self‑reported osteoarthritis, with a fully adjusted odds ratio of 2.01 (95% CI: 1.23–3.30) after controlling for demographic characteristics, lifestyle factors, metabolic indicators, and comorbidities. This finding suggests a relatively strong association between recent gastrointestinal illness and prevalent osteoarthritis in a general U.S. population. Longitudinal evidence focusing specifically on symptomatic knee OA further supports this association. In a nationwide cohort study based on the CHARLS, Ren et al. found that participants with a history of digestive diseases had a significantly increased risk of incident symptomatic knee OA during follow‑up (OR = 1.54, 95% CI: 1.30–1.82), even after adjustment for multiple sociodemographic factors, lifestyle behaviors, and chronic conditions [11]. Moreover, the cumulative incidence of symptomatic knee OA was higher among individuals with digestive diseases than among those without, suggesting that digestive system disorders may precede the development of knee‑specific osteoarthritis. The magnitude and direction of this association are broadly consistent with our findings. However, evidence from genetic causal inference studies is less consistent. In a bidirectional Mendelian randomization analysis, Xu et al. found no evidence supporting a causal effect of genetically predicted gastrointestinal disorders on osteoarthritis risk [37]. Instead, the study suggested a potential causal pathway in the opposite direction, whereby osteoarthritis increased the risk of gastroesophageal reflux disease, partly mediated by medication use. This discrepancy highlights important methodological differences between observational and genetic approaches, particularly with respect to exposure definition, disease chronicity, and the role of treatment‑related factors. Compared with previous studies, our study employed a longitudinal design with a median follow‑up of 84 months, enabling the assessment of incident symptomatic knee OA and supporting a temporal sequence between digestive system diseases and osteoarthritis onset. Moreover, by applying multiple complementary analytical strategies — including multivariable regression, propensity score matching, and inverse probability weighting — we were able to test the robustness of the association across different modeling frameworks. These methodological advantages may partly explain the slightly attenuated but more stable effect estimates observed in our study and provide more rigorous population‑based evidence extending previous findings. Although our study was not designed to directly investigate biological mechanisms, accumulating evidence supports several plausible pathways linking digestive system diseases to symptomatic knee OA. One of the most widely discussed mechanisms involves alterations in gut microbiota composition and increased intestinal permeability, often described as the “gut–joint axis” [38–40]. Digestive diseases are frequently accompanied by gut dysbiosis, which may facilitate the translocation of microbial products into systemic circulation and promote chronic low‑grade inflammation. Such inflammatory processes have been implicated in cartilage degradation, synovial inflammation, and osteoarthritis progression. In addition, digestive system diseases have been associated with metabolic dysfunction, immune dysregulation, and impaired nutrient absorption, all of which may indirectly influence musculoskeletal health. Chronic inflammation and metabolic disturbances may increase cartilage vulnerability, while deficiencies in essential nutrients could adversely affect muscle mass and joint support. Furthermore, prior studies have demonstrated close links between digestive diseases, frailty, and sarcopenia—conditions characterized by reduced muscle strength and functional decline—which may increase biomechanical stress on the knee joint and predispose individuals to osteoarthritis [14, 15]. Together, these interconnected pathways provide biological plausibility for the epidemiological association observed in our study without implying a direct causal effect. Our study is characterized by several advantages. First, based on a large, nationally representative longitudinal cohort of Chinese middle-aged and older adults with long-term prospective follow-up, we identified a higher cumulative incidence of symptomatic knee OA among individuals with digestive diseases, who may constitute a high-risk subgroup worthy of early screening and integrated management. Second, we specifically targeted digestive diseases as the key exposure, allowing for a focused and in-depth analysis of its association with incident symptomatic knee OA. Third, we applied PSM and other complementary analytical strategies, including sensitivity analyses for reverse causation, competing risk, and unmeasured confounding, to enhance the reliability and robustness of our findings while minimizing potential confounding and selection bias. Nevertheless, several limitations should be acknowledged. First, due to lack of ICD codes and subtype data in CHARLS, we were unable to distinguish between specific digestive diseases such as inflammatory bowel disease and functional dyspepsia or evaluate disease severity, which may obscure potential differences in their association with symptomatic knee OA risk. Second, symptomatic knee OA was defined based on self-reported physician diagnosis and knee pain, which may be subject to misclassification as other forms of joint pain or arthropathy. Third, data on NSAID use were unavailable in the CHARLS dataset, and we did not adjust for important socioeconomic dimensions such as income and healthcare access, both of which may represent sources of residual confounding. Future studies with dedicated socioeconomic measures and more complete data should further explore the potential role of these factors. Fourth, with respect to population representativeness, our study cohort tended to be relatively older, female-dominant, and metabolically disadvantaged. The exclusion of participants with missing covariate data may introduce selection bias, which should be considered when extrapolating our results to other populations. From a clinical and public health perspective, these results underscore the importance of considering symptomatic knee OA as a condition influenced by systemic health rather than an isolated joint disorder. Individuals with digestive system diseases may represent a subgroup at elevated risk for symptomatic knee OA and could benefit from early screening and integrated management strategies addressing metabolic health, physical function, and psychological well‑being. Conversely, patients presenting with symptomatic knee OA may warrant assessment for coexisting digestive and cardiometabolic conditions to optimize overall outcomes. Conclusion In conclusion, this longitudinal study demonstrates that digestive system diseases are associated with an increased risk of incident symptomatic knee OA. These findings extend previous cross‑sectional and longitudinal evidence by supporting a temporal relationship between digestive disorders and symptomatic knee OA and highlight digestive diseases as a potential component of the broader multimorbidity network influencing musculoskeletal health. Future studies are warranted to clarify the underlying biological mechanisms and to explore whether improved management of digestive system diseases may contribute to the prevention or delayed onset of symptomatic knee OA. Supplementary Information Supplementary Material 1. Supplementary Material 2. Acknowledgements We thank Peking University for providing public access to CHARLS. We also acknowledge Jie Liu (Department of Vascular and Endovascular Surgery, Chinese PLA General Hospital & Physician-Scientist Center of China) for his insightful input. Abbreviations Osteoarthritis China Health and Retirement Longitudinal Study Standard Deviation Interquartile Range Propensity score matching Standardized Mean Differences Body Mass Index Metabolic Equivalent of Task High-density lipoprotein Hazard Ratio Confidence Intervals Authors’ contributions ShC, LJ, and MZ developed the methodological framework and study design. ShC conducted the data analysis, prepared all figures and tables, and drafted the initial manuscript. SuC, AL and BW participated in manuscript revision. DZ, WX, and YL contributed to the refinement of the figures. All authors contributed to critical revision and approved the final manuscript. Funding No external funding was provided for this research. Data availability The data analyzed in this study are publicly available through online repositories. Information regarding the relevant repository names and corresponding accession numbers is provided below. Specifically, the datasets used in this analysis can be obtained from the official CHARLS website (https://charls.pku.edu.cn/en/). Declarations Ethics approval and consent to participate The CHARLS was implemented in accordance with the Declaration of Helsinki and received approval from the Institutional Review Board (IRB) of Peking University (IRB00001052-11015). All participants provided written informed consent prior to their enrollment in the study. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Shile Chen contributed as first author. Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Supplementary Materials Supplementary Material 1. Supplementary Material 2. Data Availability Statement The data analyzed in this study are publicly available through online repositories. Information regarding the relevant repository names and corresponding accession numbers is provided below. Specifically, the datasets used in this analysis can be obtained from the official CHARLS website (https://charls.pku.edu.cn/en/).