Abstract
Background
The American Heart Association (AHA) introduced the Cardiovascular-Kidney-Metabolic (CKM) syndrome framework, emphasizing the interplay of metabolic risk factors, chronic kidney disease, and cardiovascular disease (CVD). Whether the metabolic syndrome (MetS) score—a simple count of metabolic abnormalities—provides additional risk information within CKM stages 0–3 remains uncertain. This study examined the association between MetS score and incident CVD in a CHARLS-based cohort with CKM stages 0–3.
Methods
We analyzed 7889 participants aged ≥45 years from the China Health and Retirement Longitudinal Study (CHARLS) with CKM stages 0–3 and no baseline CVD. CKM staging was corrected to enforce hierarchical consistency (participants with ≥2 MetS components were classified as at least Stage 2). MetS score (0–5) was defined per harmonized criteria. The primary outcome was incident CVD (self-reported heart disease or stroke) over 7-year follow-up. Cox proportional hazards models (with CKM stage in the primary model, without adjusting for MetS components), restricted cubic spline analyses, Modified Poisson regression, subgroup analyses, and sensitivity analyses were performed. Single iterative imputation addressed missing covariates (<3% per variable).
Results
During median 7.0-year follow-up, 815 participants (10.3%) developed CVD. After full adjustment including CKM stage, each 1-point higher MetS score was associated with 22.3% higher CVD risk (HR: 1.223, 95% CI: 1.143–1.308, P < 0.001). The highest quartile had over twice the risk of the lowest (HR: 2.206, 95% CI: 1.625–2.996, P < 0.001; P for trend < 0.001). A categorical dose-response was confirmed (score 5 vs. 0: HR 3.379). The association was stronger for stroke (HR: 1.333) than heart disease (HR: 1.119; P for heterogeneity = 0.009). Adding MetS score to CKM stage significantly improved model fit (likelihood-ratio P < 0.001) but yielded modest discrimination improvement (ΔC-statistic: 0.017, 95% CI: 0.006–0.029). Results were robust across sensitivity analyses.
Conclusions
In this CHARLS-based cohort of Chinese adults with CKM stages 0–3, a higher MetS score was independently associated with elevated CVD risk in a graded, dose-response manner. The MetS score captures cumulative metabolic burden beyond CKM staging, though its incremental discriminative value is modest. Further validation with adjudicated outcomes and accurate event timing is warranted.
Keywords: Cardiovascular-kidney-metabolic syndrome, Metabolic syndrome, Cardiovascular disease, Cohort study, CHARLS
1. Introduction
Cardiovascular disease (CVD) remains the leading cause of mortality worldwide, accounting for an estimated 19.8 million deaths in 2022 [1]. The pathophysiology of CVD is increasingly understood as a systemic process involving the interplay of metabolic dysregulation, renal dysfunction, and cardiovascular pathology [2,3]. In October 2023, the American Heart Association (AHA) formally introduced the Cardiovascular-Kidney-Metabolic (CKM) syndrome framework in a landmark Presidential Advisory [4]. This framework categorizes individuals along a continuum from stage 0 (no CKM risk factors) through stage 4 (established clinical CVD), recognizing that metabolic risk factors, chronic kidney disease (CKD), and subclinical cardiovascular pathology interact synergistically to accelerate disease progression [4].
The CKM framework has profound implications for cardiovascular prevention, as it emphasizes early identification and intervention in the preclinical stages (stages 0–3) before the onset of overt CVD [4,5]. Within this context, identifying reliable, cost-effective, and clinically accessible risk markers for CVD among individuals in CKM stages 0–3 has emerged as a research priority [6,7].
Metabolic syndrome (MetS) represents a clustering of cardiometabolic risk factors—including abdominal obesity, dyslipidemia, hypertension, and hyperglycemia—that collectively confer substantially elevated CVD risk [8,9]. The MetS score, a continuous count of positive MetS components (range: 0–5), provides a graded measure of metabolic dysregulation [10]. Previous studies have demonstrated that the MetS score is associated with incident CVD in the general population [11,12] and in specific subpopulations [13,14]. However, the association between MetS score and CVD specifically within the CKM framework—where metabolic risk factors are already incorporated into staging—has not been examined.
This knowledge gap is clinically important for two reasons. First, within each CKM stage, substantial heterogeneity in metabolic burden exists; the MetS score may capture residual risk beyond what is conveyed by categorical staging alone. Second, because CKM Stage 2 already incorporates metabolic risk factors, any analysis of MetS within CKM stages must be framed as assessing cumulative or residual metabolic burden rather than an entirely independent pathway.
Therefore, using data from the China Health and Retirement Longitudinal Study (CHARLS), a large prospective cohort of Chinese adults aged 45 years and older, the present study aimed to investigate the prospective association between MetS score and incident CVD in a population with CKM stages 0–3.
2. Methods
2.1. Data source and study population
The CHARLS is a large longitudinal cohort study that surveys Chinese community-dwelling adults aged 45 years and older. The study employed a multi-stage stratified probability-proportional-to-size sampling strategy, enrolling participants from 450 villages or communities across 28 provinces [15]. The CHARLS was conducted in accordance with the Declaration of Helsinki and received ethical approval from the Institutional Review Board at Peking University (IRB00001052-11015). All participants provided written informed consent.
The present study utilized data from the 2011–2012 baseline wave (Wave 1), with follow-up assessments in 2013 (Wave 2), 2015 (Wave 3), and 2018 (Wave 4). The present analyses did not incorporate CHARLS sampling weights, strata, or primary sampling units. Accordingly, the results should be interpreted as associations within the CHARLS analytic sample rather than as nationally representative effect estimates.
2.2. Inclusion and exclusion criteria
Participants were included if they: (1) were aged ≥45 years at baseline; (2) completed the baseline survey and blood biomarker assessment; (3) had complete data for MetS components; (4) were classified as CKM stages 0–3; and (5) had at least one follow-up visit. Participants were excluded if they had a history of CVD at baseline, were missing baseline CVD status, had CKM stage 4, or lacked follow-up data. A detailed flowchart is presented in Fig. 1.
Fig. 1.
Participant selection flowchart from the CHARLS 2011 baseline survey. CVD, cardiovascular disease; MetS, metabolic syndrome; CKM, cardiovascular-kidney-metabolic syndrome.
2.3. Definition of metabolic syndrome score
The MetS score was defined according to harmonized criteria [8] with Asian-specific modifications [16]. One point was assigned for each of: (1) waist circumference ≥90 cm (men) or ≥80 cm (women); (2) triglycerides ≥150 mg/dL; (3) HDL-C <40 mg/dL (men) or <50 mg/dL (women); (4) blood pressure ≥130/85 mmHg or antihypertensive medication; (5) fasting glucose ≥100 mg/dL or glucose-lowering medication. The score ranged from 0 to 5.
2.4. Definition of CKM stages (corrected)
CKM stages 0–3 were defined per the AHA Presidential Advisory [4] with a corrected hierarchical algorithm. Stage 0: absence of excess adiposity, metabolic risk factors, and subclinical CVD. Stage 1: excess adiposity (BMI ≥25 kg/m2 or elevated waist circumference) without other metabolic risk factors. Stage 2: presence of metabolic risk factors (MetS, hypertension, diabetes, hypertriglyceridemia, or CKD) without subclinical CVD. Stage 3: subclinical CVD (2008 Framingham General CVD Risk Score (D'Agostino et al. Circulation 2008) ≥20% or eGFR <30 mL/min/1.73 m2). eGFR was estimated using the Chinese Modification of Diet in Renal Disease (C-MDRD) equation [17].
Importantly, the hierarchical algorithm was corrected to enforce logical consistency: any participant with ≥2 MetS components was classified as at least Stage 2, regardless of adiposity status. In the original algorithm, 530 participants were reclassified (182 as Stage 0 and 331 as Stage 1 despite having MetS score ≥2); an additional 17 Stage 1 participants whose single MetS component was not waist circumference were also reclassified. The corrected distribution was: Stage 0 (n = 1620), Stage 1 (n = 235), Stage 2 (n = 5143), Stage 3 (n = 891).
2.5. Outcome ascertainment
CVD events were ascertained at each follow-up wave (2013, 2015, 2018) via self-reported physician-diagnosed heart disease or stroke. Because CHARLS collects diagnoses at discrete wave intervals rather than through continuous surveillance, event times are interval-censored. In the analytic file, the vast majority of incident CVD events (798 of 815, 97.9%) were first reported at the 2018 wave; events detected at earlier waves were assigned to the midpoint of the corresponding interval. This wave-level ascertainment provides approximate event timing while acknowledging that precise event dates were unavailable.
2.6. Statistical analysis
Baseline characteristics were described by MetS score quartiles (Q1: 0–1, Q2: 2, Q3: 3, Q4: 4–5). Cox proportional hazards regression estimated hazard ratios (HRs) with progressive adjustment: Model 1 (unadjusted); Model 2 (age, sex); Model 3 (+ marital status, education, smoking, drinking); Model 4 (+BUN, creatinine, TC, LDL-C, CRP, uric acid, platelet count; HDL-C was excluded as a MetS component to avoid overadjustment); Model 5 (primary: + CKM stage as categorical variable; hypertension and diabetes were excluded as MetS components). The original Models 4 and 5 (including HDL-C, hypertension, and diabetes) are presented as supplementary sensitivity analyses.
The proportional hazards assumption was tested using Schoenfeld residuals. The MetS score violated the PH assumption (P = 0.003); accordingly, results should be interpreted as average hazard ratios over the follow-up period, and a Modified Poisson regression with robust standard errors was conducted as a complementary analysis for 7-year cumulative risk. Restricted cubic spline analysis with 3 knots (10th, 50th, 90th percentiles) assessed non-linearity; a quadratic term test was also performed. A categorical (ordinal) analysis treating each MetS score value (0–5) as a separate indicator was conducted as the primary dose-response assessment.
Subgroup analyses were conducted by age, sex, smoking, CKM stage, diabetes, and CKD status, with formal interaction tests. CKM-stratified analyses examined the association within CKM stage groups. For CKM Stage 3, progressive adjustment (age only; age + sex; age + sex + smoking + drinking) was performed due to the limited number of events (n = 120) and extreme sex imbalance (94.6% male). Heart disease and stroke were analyzed as separate outcomes with a formal heterogeneity test.
Incremental predictive value was assessed by comparing CKM stage alone versus CKM stage + MetS score using Harrell's C-statistic (with bootstrap 95% CI), likelihood-ratio test, net reclassification improvement (NRI), and integrated discrimination improvement (IDI).
Sensitivity analyses included: (1) complete-case analysis; (2) binary MetS (≥3 vs. <3); (3) excluding CKM Stage 0; (4) alternative event-time assignment (beginning, midpoint, and end of interval); (5) Modified Poisson regression for 7-year risk. Missing covariates in the candidate covariate set were sparse (<3% per variable) and were initially handled using single iterative imputation (Bayesian Ridge regression) via scikit-learn's IterativeImputer with 10 iterations, which produces a single completed dataset but does not propagate imputation uncertainty. After application of the inclusion criteria, however, the final analytic sample (N = 7889) contained no missing values on any covariate entered in the primary model; the only variable with residual missingness was eGFR (missing for 3 participants), which was used solely to define CKD status and was not a model covariate. Because imputation therefore did not change the analytic dataset, the complete-case sensitivity analysis was necessarily performed on the identical 7889 participants with the same number of events, yielding the same estimate.
All analyses used Python 3.11 (lifelines 0.30.0, scikit-learn, statsmodels). Two-sided P < 0.05 was considered significant.
3. Results
3.1. Baseline characteristics
A total of 7889 participants were included (Fig. 1). The mean age was 59.1 ± 9.3 years, and 4163 (52.8%) were female. The corrected CKM stage distribution was: Stage 0 (n = 1,620, 20.5%), Stage 1 (n = 235, 3.0%), Stage 2 (n = 5,143, 65.2%), Stage 3 (n = 891, 11.3%). The MetS score distribution was: 0 (n = 892, 11.3%), 1 (n = 1,820, 23.1%), 2 (n = 1,998, 25.3%), 3 (n = 1,609, 20.4%), 4 (n = 1,018, 12.9%), 5 (n = 552, 7.0%). The Spearman correlation between MetS score and CKM stage was rho = 0.611 (P < 0.001).
Baseline characteristics by MetS quartiles are presented in Table 1, and the distribution of metabolic burden and CKM stages is summarized visually in Fig. 2. Higher quartiles had progressively higher BMI, waist circumference, blood pressure, triglycerides, fasting glucose, and prevalence of hypertension and diabetes, with lower HDL-C (all P < 0.01). CKM Stages 0 and 1 were confined to Q1, consistent with the corrected hierarchical staging.
Table 1.
Baseline characteristics by MetS score quartile.
| Characteristic | Q1 (n = 2712) | Q2 (n = 1998) | Q3 (n = 1609) | Q4 (n = 1570) | P |
|---|---|---|---|---|---|
| Age, years | 58.4 ± 9.1 | 59.3 ± 9.5 | 59.7 ± 9.4 | 59.2 ± 9.0 | <0.001 |
| Female, n (%) | 1061 (39.1) | 1007 (50.4) | 993 (61.7) | 1102 (70.2) | <0.001 |
| BMI, kg/m2 | 21.4 ± 2.8 | 22.9 ± 3.4 | 24.7 ± 3.8 | 26.1 ± 3.5 | <0.001 |
| Waist, cm | 78.7 ± 7.1 | 83.9 ± 8.9 | 89.1 ± 8.8 | 92.8 ± 8.5 | <0.001 |
| SBP, mmHg | 119.4 ± 16.7 | 131.5 ± 20.9 | 136.0 ± 21.0 | 141.3 ± 21.4 | <0.001 |
| DBP, mmHg | 70.3 ± 10.7 | 76.4 ± 11.3 | 78.5 ± 11.9 | 81.3 ± 11.9 | <0.001 |
| TG, mg/dL | 84.7 ± 32.1 | 105.1 ± 49.7 | 143.0 ± 91.8 | 239.3 ± 173.6 | <0.001 |
| HDL-C, mg/dL | 59.3 ± 14.1 | 54.5 ± 14.7 | 47.6 ± 13.2 | 38.2 ± 9.0 | <0.001 |
| FBG, mg/dL | 97.2 ± 18.6 | 108.5 ± 32.6 | 114.9 ± 39.4 | 127.9 ± 47.7 | <0.001 |
| eGFR, mL/min | 110.6 ± 26.8 | 108.9 ± 28.7 | 107.0 ± 27.8 | 106.7 ± 33.0 | <0.001 |
| Hypertension, n (%) | 537 (19.8) | 1130 (56.6) | 1088 (67.6) | 1315 (83.8) | <0.001 |
| Diabetes, n (%) | 141 (5.2) | 276 (13.8) | 331 (20.6) | 521 (33.2) | <0.001 |
| MetS score | 0.7 ± 0.5 | 2.0 ± 0.0 | 3.0 ± 0.0 | 4.4 ± 0.5 | <0.001 |
| CKM Stage 0, n (%) | 1620 (59.7) | 0 (0.0) | 0 (0.0) | 0 (0.0) | <0.001 |
| CKM Stage 1, n (%) | 235 (8.7) | 0 (0.0) | 0 (0.0) | 0 (0.0) | <0.001 |
| CKM Stage 2, n (%) | 737 (27.2) | 1727 (86.4) | 1384 (86.0) | 1295 (82.5) | <0.001 |
| CKM Stage 3, n (%) | 120 (4.4) | 271 (13.6) | 225 (14.0) | 275 (17.5) | <0.001 |
Values are mean ± SD or n (%). P values: ANOVA for continuous, χ2 for categorical. CKM stages 0–1 are confined to Q1 after hierarchical staging correction.
Fig. 2.
Baseline metabolic burden. (A) Distribution of MetS score (0–5) with observed CVD incidence rate. (B) CKM stage distribution across MetS score quartiles (corrected hierarchical staging).
3.2. Association between MetS score and incident CVD
During median 7.0-year follow-up, 815 participants (10.3%) developed incident CVD, comprising 489 heart disease events and 378 stroke events; 52 participants experienced both conditions, accounting for the overlap in component counts.
Table 2 presents the Cox regression results, with key estimates visualized in Fig. 3. In the primary fully adjusted model (Model 5, including CKM stage but excluding MetS components), each 1-point higher MetS score was associated with 22.3% higher CVD risk (HR: 1.223, 95% CI: 1.143–1.308, P < 0.001). The association was stronger than in the original overadjusted model (HR: 1.164), confirming that adjusting for MetS components attenuated the estimate.
Table 2.
Cox models for per 1-point MetS score increase and incident CVD.
| Model | N | Events | HR | 95% CI | P |
|---|---|---|---|---|---|
| Model 1: Unadjusted | 7889 | 815 | 1.234 | 1.176–1.294 | <0.001 |
| Model 2: Age + sex | 7889 | 815 | 1.224 | 1.165–1.286 | <0.001 |
| Model 3: + Demographics | 7889 | 815 | 1.220 | 1.161–1.282 | <0.001 |
| Model 4: + Biomarkers (no HDL-C) | 7889 | 815 | 1.227 | 1.165–1.294 | <0.001 |
| Model 5: + CKM stage (primary) | 7889 | 815 | 1.223 | 1.143–1.308 | <0.001 |
| Model 4-orig: + HDL-C | 7889 | 815 | 1.265 | 1.185–1.351 | <0.001 |
| Model 5-orig: + HTN + DM | 7889 | 815 | 1.164 | 1.074–1.262 | <0.001 |
Model 5 (primary) adjusts for age, sex, marital status, education, smoking, drinking, BUN, creatinine, TC, LDL-C, CRP, uric acid, platelet count, and CKM stage. HDL-C, hypertension, and diabetes are excluded as MetS components. Original models (4-orig, 5-orig) shown for comparison.
Fig. 3.
Forest plot of the association between MetS score and incident CVD. Left: continuous MetS score (per 1-point increase) across progressive adjustment models. Right: quartile analysis (Q2–Q4 vs. Q1) in the primary model. Squares indicate HRs; horizontal lines indicate 95% CIs.
Quartile analysis in the primary model showed a pronounced dose-response gradient: Q2 vs. Q1 HR 1.480 (95% CI: 1.097–1.998), Q3 vs. Q1 HR 1.822 (95% CI: 1.346–2.468), Q4 vs. Q1 HR 2.206 (95% CI: 1.625–2.996; P for trend < 0.001; Table 3).
Table 3.
Fully adjusted association between MetS score quartiles and incident CVD (primary model).
| Quartile | N | Events | HR | 95% CI | P |
|---|---|---|---|---|---|
| Q1 (0–1) | 2712 | 193 | 1.000 (Ref) | – | – |
| Q2 (2) | 1998 | 194 | 1.480 | 1.097–1.998 | 0.010 |
| Q3 (3) | 1609 | 194 | 1.822 | 1.346–2.468 | <0.001 |
| Q4 (4–5) | 1570 | 234 | 2.206 | 1.625–2.996 | <0.001 |
| P for trend | <0.001 |
Primary model includes CKM stage. Q1 is reference. P for trend estimated by modeling quartile order as continuous.
3.3. Dose-response analysis
Categorical analysis treating each MetS score value as a separate indicator confirmed a monotonic dose-response: compared with score 0, HRs were 1.495 (score 1), 2.154 (score 2), 2.657 (score 3), 3.148 (score 4), and 3.379 (score 5; all P < 0.05 except score 1 P = 0.028; Fig. 4). The quadratic non-linearity test was borderline significant (P = 0.049), suggesting slight attenuation at the highest scores, though the overall pattern was predominantly linear. The observed cumulative CVD incidence by MetS quartile over 7-year follow-up is shown in Fig. 5.
Fig. 4.
Dose-response relationship between MetS score (0–5) and incident CVD. Categorical analysis with score 0 as reference. Shaded region: 95% CI. Dashed line: HR = 1.0.
Fig. 5.
Observed cumulative incidence of CVD (events/N per quartile) by MetS score quartile over 7-year follow-up. Note: these are observed event proportions, not Kaplan-Meier estimates; the competing risk of death is not accounted for. Q1 (score 0–1): 193/2712; Q2 (score 2): 194/1998; Q3 (score 3): 194/1609; Q4 (score 4–5): 234/1570.
3.4. Subgroup and interaction analysis
The positive association was consistent across subgroups (Table 4, Fig. 6). No significant interactions were detected for age (P = 0.087), sex (P = 0.815), or CKM stage (P = 0.188). The association was significant in CKM Stage 2 (HR: 1.196, P < 0.001) but not in Stage 0 (HR: 1.410, P = 0.088) or Stage 3 (HR: 1.162, P = 0.105), likely reflecting limited power in these smaller subgroups.
Table 4.
Subgroup analysis of the association between MetS score and incident CVD.
| Subgroup | N | Events | HR | 95% CI | P value |
|---|---|---|---|---|---|
| Age <60 years | 4433 | 416 | 1.249 | 1.135–1.375 | <0.001 |
| Age ≥60 years | 3456 | 399 | 1.168 | 1.060–1.286 | 0.002 |
| Male | 3726 | 345 | 1.279 | 1.147–1.426 | <0.001 |
| Female | 4163 | 470 | 1.165 | 1.066–1.273 | <0.001 |
| Never smoker | 4796 | 513 | 1.196 | 1.098–1.302 | <0.001 |
| Former smoker | 647 | 77 | 1.332 | 1.052–1.686 | 0.017 |
| Current smoker | 2446 | 225 | 1.229 | 1.078–1.401 | 0.002 |
| CKM Stage 0 | 1620 | 105 | 1.410 | 0.950–2.091 | 0.088 |
| CKM Stage 2 | 5143 | 558 | 1.196 | 1.107–1.291 | <0.001 |
| CKM Stage 3 | 891 | 120 | 1.162 | 0.969–1.392 | 0.105 |
HR, hazard ratio; CI, confidence interval. Per 1-point increase in MetS score, adjusted for age, sex, marital status, education, smoking, drinking, BUN, creatinine, TC, LDL-C, CRP, uric acid, and platelet count. CKM Stage 1 (n = 235) not shown due to model convergence limitations. P for interaction: age = 0.087, sex = 0.815, CKM stage = 0.188.
Fig. 6.
Subgroup forest plot for the association between MetS score (per 1-point) and incident CVD. P values for interaction: age 0.087, sex 0.815, CKM stage 0.188. In the diabetes subgroup, the association was stronger among participants with diabetes (HR: 1.439, 95% CI: 1.251–1.656) than without (HR: 1.172, 95% CI: 1.104–1.244; P for interaction reported in supplementary material).
3.5. CKM-stratified analysis
In the combined CKM 0–1 group, the association was significant (HR: 1.661, 95% CI: 1.154–2.392, P = 0.006). In CKM 2–3, the association was stronger and highly significant (HR: 1.232, 95% CI: 1.151–1.319, P < 0.001; Table 5).
Table 5.
Association between MetS score and incident CVD within CKM strata.
| CKM stratum | N | Events | HR | 95% CI | P value |
|---|---|---|---|---|---|
| CKM stages 0–1 | 1855 | 137 | 1.661 | 1.154–2.392 | 0.006 |
| CKM stages 2–3 | 6034 | 678 | 1.232 | 1.151–1.319 | <0.001 |
| CKM stages 1–3 | 6269 | 710 | 1.196 | 1.121–1.276 | <0.001 |
| CKM stage 3 only (age only) | 891 | 120 | 1.232 | 1.057–1.436 | 0.008 |
| CKM stage 3 only (age + sex) | 891 | 120 | 1.196 | 1.019–1.404 | 0.029 |
| CKM stage 3 only (age + sex + smoking + drinking) | 891 | 120 | 1.184 | 1.002–1.399 | 0.048 |
HR, hazard ratio; CI, confidence interval. CKM, cardiovascular-kidney-metabolic. Within-stratum models do not adjust for CKM stage.
For CKM Stage 3, progressive adjustment yielded: age-only HR 1.232 (P = 0.008); age + sex HR 1.196 (P = 0.029); age + sex + smoking + drinking HR 1.184 (P = 0.048). CKM Stage 3 was defined partly using the Framingham Risk Score, which shares components with both the MetS score and the covariate set; these findings should be interpreted with caution and are not directly comparable to fully adjusted estimates from other stages.
3.6. Heart disease and stroke
MetS score was associated with both endpoints, with a significantly stronger association for stroke (HR: 1.333, 95% CI: 1.209–1.470) than heart disease (HR: 1.119, 95% CI: 1.025–1.221; P for heterogeneity = 0.009; Table 6). The quartile gradient for stroke was particularly pronounced (Q4 vs. Q1: HR 3.367, 95% CI: 2.075–5.463).
Table 6.
Association of MetS score with incident heart disease and stroke.
| Outcome/exposure | N | Events | HR | 95% CI | P value |
|---|---|---|---|---|---|
| Heart disease: per 1-point MetS | 7889 | 489 | 1.119 | 1.025–1.221 | 0.012 |
| Heart disease: Q2 vs Q1 | 7889 | 489 | 1.042 | 0.728–1.492 | 0.822 |
| Heart disease: Q3 vs Q1 | 7889 | 489 | 1.296 | 0.903–1.861 | 0.160 |
| Heart disease: Q4 vs Q1 | 7889 | 489 | 1.444 | 1.001–2.084 | 0.050 |
| Stroke: per 1-point MetS | 7889 | 378 | 1.333 | 1.209–1.470 | <0.001 |
| Stroke: Q2 vs Q1 | 7889 | 378 | 2.103 | 1.308–3.380 | 0.002 |
| Stroke: Q3 vs Q1 | 7889 | 378 | 2.420 | 1.493–3.924 | <0.001 |
| Stroke: Q4 vs Q1 | 7889 | 378 | 3.367 | 2.075–5.463 | <0.001 |
HR, hazard ratio; CI, confidence interval. Q1 is the reference quartile. P for heterogeneity between heart disease and stroke associations = 0.009.
3.7. Incremental predictive value
Adding MetS score to CKM stage significantly improved model fit (likelihood-ratio χ2 = 34.37, P < 0.001). The improvement in discrimination, although statistically significant, was modest: C-statistic increased from 0.607 (CKM + covariates) to 0.624 (CKM + MetS + covariates), a difference of 0.017 (bootstrap 95% CI 0.006 to 0.029, which excludes zero). The IDI was 0.006 and the category-free NRI was 0.222. These results indicate that while MetS score is significantly associated with CVD risk beyond CKM stage, its incremental discriminative value for individual risk prediction is limited.
3.8. Sensitivity analyses
All sensitivity analyses confirmed robustness (Table 7, Fig. 7): complete-case analysis (identical to the analytic sample, N = 7889; HR: 1.223), binary MetS ≥3 vs. <3 (HR: 1.482, 95% CI: 1.258–1.745), excluding CKM Stage 0 (HR: 1.217, 95% CI: 1.136–1.304), and alternative event-time assignment (HR: 1.223). Modified Poisson regression for 7-year cumulative risk yielded consistent results (RR: 1.205, 95% CI: 1.135–1.280; Q4 vs. Q1 RR: 2.090, 95% CI: 1.565–2.793).
Table 7.
Sensitivity analyses.
| Sensitivity analysis | N | Events | HR/RR | 95% CI | P |
|---|---|---|---|---|---|
| Primary (Model 5) | 7889 | 815 | 1.223 | 1.143–1.308 | <0.001 |
| Complete case | 7889 | 815 | 1.223 | 1.143–1.308 | <0.001 |
| Binary MetS (≥3 vs. <3) | 7889 | 815 | 1.482 | 1.258–1.745 | <0.001 |
| Excluding CKM Stage 0 | 6269 | 710 | 1.217 | 1.136–1.304 | <0.001 |
| Events at interval beginning | 7889 | 815 | 1.223 | 1.143–1.308 | <0.001 |
| Modified Poisson (7-yr RR) | 6779 | 815 | 1.205 | 1.135–1.280 | <0.001 |
| Heart disease | 7889 | 489 | 1.119 | 1.025–1.221 | 0.012 |
| Stroke | 7889 | 378 | 1.333 | 1.209–1.470 | <0.001 |
HR from Cox regression unless otherwise noted. Modified Poisson reports risk ratio (RR) with robust SE for 7-year cumulative risk among participants with 7-year follow-up.
Fig. 7.
Sensitivity analysis forest plot. HR, hazard ratio; RR, risk ratio (Modified Poisson). All models use the primary covariate set unless otherwise noted.
4. Discussion
In this prospective cohort study of Chinese adults aged 45 years and older with CKM stages 0–3, we found that a higher MetS score was independently associated with elevated CVD risk in a graded, dose-response manner. Each 1-point increase in MetS score was associated with a 22% higher CVD risk after adjustment for CKM stage and covariates excluding MetS components. The highest quartile had over twice the risk of the lowest. The association was consistent across subgroups, robust to multiple sensitivity analyses, and confirmed by Modified Poisson regression.
4.1. Comparison with previous studies
Previous studies have demonstrated the prognostic value of MetS for CVD in the general population [11,12,18] and in high-risk groups [19,20]. However, this is among the first studies to examine the MetS-CVD association specifically within the CKM framework. Our finding that MetS score retains a significant association after adjustment for CKM stage suggests that the continuous score captures a dimension of cumulative metabolic burden that categorical staging does not fully represent. However, the modest improvement in discrimination (ΔC-statistic: 0.017) indicates that this additional information, while statistically significant, has limited incremental value for individual risk prediction.
The dose-response gradient (score 5 vs. 0: HR 3.38) reflects the incremental cardiovascular risk associated with each additional MetS component [21], and is comparable in magnitude to estimates from previous cohort studies [22,23]. The borderline non-linearity (P = 0.049) suggests slight attenuation at the highest scores, possibly reflecting a ceiling effect of metabolic burden.
4.2. Differential associations with heart disease and stroke
The significantly stronger association with stroke (HR: 1.333) than heart disease (HR: 1.119; P for heterogeneity = 0.009) is consistent with evidence that metabolic factors are especially potent cerebrovascular risk factors [24,25]. The MetS components collectively promote a pro-inflammatory, pro-oxidative, and pro-thrombotic state that accelerates atherosclerosis and endothelial dysfunction [26]; abdominal adiposity drives systemic inflammation and insulin resistance [27], hyperglycemia promotes advanced glycation end-product formation and oxidative stress [28], dyslipidemia generates highly atherogenic small, dense LDL particles [29], and hypertension exerts mechanical stress on the vascular endothelium [30]. Declining renal function further magnifies these cardiovascular consequences [31]. These pathways are particularly consequential in East Asian populations, where stroke accounts for a disproportionately large share of the CVD burden [32]. This may partly reflect differential misclassification: stroke is a more acute and clinically dramatic event, less likely to be under-reported than the heterogeneous category of “heart disease."
These findings should be interpreted within the rapidly expanding cardiovascular-kidney-metabolic (CKM) literature. Recent large cohort studies have reported graded increases in atherosclerotic and total cardiovascular risk across successive CKM stages [33], reinforcing the staging framework as a continuum of accumulating cardiometabolic burden. Beyond atherosclerotic outcomes, CKM staging has also been linked to mental-health comorbidity, including a stage-dependent increase in depression risk [34], and depressive symptoms have been shown to accelerate cardiovascular disease progression across CKM stages 0–3 [35]. Collectively, these studies indicate that CKM stages capture a broad spectrum of multi-system risk. Our results complement this literature by demonstrating that, even after accounting for CKM stage, a continuous metabolic syndrome score retains an independent, dose-dependent association with incident CVD—particularly stroke—although its incremental discriminative value for individual-level risk prediction remains limited.
4.3. CKM stage 3 findings
The association within CKM Stage 3 (age-adjusted HR: 1.232, P = 0.008) persisted with progressive adjustment but should be interpreted cautiously. Stage 3 was defined partly using the Framingham Risk Score, which incorporates age, sex, smoking, blood pressure, cholesterol, and diabetes—variables that are also included in the MetS score and the covariate set. Furthermore, the Framingham Risk Score has not been recalibrated for Chinese populations, which may introduce additional measurement error in the Stage 3 classification. These factors introduce collinearity and potential selection bias, and the Stage 3 findings should be interpreted with caution and are not directly comparable to fully adjusted estimates from other stages. The original text read: variables that overlap with both the MetS score and the covariate set. Conditioning on this derived variable may introduce collinearity and selection bias. Of the 891 Stage 3 participants, only 7 met the eGFR < 30 criterion; the remainder were classified based on FRS ≥ 20%. The Stage 3 findings are not directly comparable to fully adjusted estimates from other stages.
4.4. Methodological considerations
Several methodological issues deserve comment. First, the proportional hazards assumption was violated for the MetS score (Schoenfeld P = 0.003), likely reflecting the interval-censored nature of event times (all events at the 7-year wave). The Modified Poisson regression, which does not require the PH assumption, yielded consistent results (RR: 1.205), supporting the robustness of the association. Because the timing of events was approximate and the PH assumption was violated, the Cox hazard ratios should be interpreted as average effects over the entire follow-up period rather than instantaneous hazard ratios; the magnitude of the reported HRs should therefore be read with caution, and the Modified Poisson estimates are regarded as the primary confirmation of the association.
Second, the original analysis adjusted for HDL-C (Model 4) and hypertension/diabetes (Model 5), which are components of the MetS score. This overadjustment attenuated the estimate from HR 1.227 to 1.164. The revised primary model excludes these variables and includes CKM stage instead, providing a more appropriate estimate of the total association between metabolic burden and CVD risk.
Third, the CKM staging algorithm was corrected to enforce hierarchical consistency. The original algorithm misclassified 530 participants (182 Stage 0 and 331 Stage 1 with MetS score ≥2 and 17 Stage 1 with a non-adiposity single MetS component) who should have been classified as Stage 2. This correction strengthened the association by reducing exposure misclassification within CKM strata.
Fourth, the MetS score has only 6 discrete values, making restricted cubic spline analysis potentially unstable. We therefore presented the categorical (ordinal) analysis as the primary dose-response assessment, with RCS as a supplementary visualization.
4.5. Strengths and limitations
Strengths include the large prospective cohort design, comprehensive covariate adjustment, multiple analytical approaches, and extensive sensitivity analyses. The corrected CKM staging and redesigned models address key methodological concerns.
Several limitations must be acknowledged. First, CVD outcomes were self-reported rather than adjudicated; although misclassification may occur, non-differential misclassification would bias estimates toward the null. Second, all incident events were ascertained at the 2018 wave, precluding true time-to-event analysis; the Cox model provides approximate average hazard ratios, and the Modified Poisson analysis serves as the primary confirmatory analysis. Third, the MetS score was assessed at a single time point; regression dilution bias may attenuate the true association. Fourth, CHARLS sampling weights were not applied; results represent associations within the analytic sample rather than nationally representative estimates. Fifth, single iterative imputation does not propagate imputation uncertainty, though the low missing proportion (<3%) limits practical impact. Sixth, the FRS-based Stage 3 definition introduces circularity when examining MetS as a CVD predictor. Seventh, explicit competing-risk data (deaths) were not available for Fine-Gray analysis. Eighth, CHARLS lacks physical activity, dietary, and detailed medication data, precluding assessment of these potential confounders and mediators. Ninth, the small number of CKD participants (n = 133) and CKM Stage 3 events (n = 120) limited subgroup power. Additionally, although baseline CVD was excluded by self-report, subclinical or undiagnosed prevalent disease may have influenced both MetS component levels and subsequent CVD reporting (reverse causation); a sensitivity analysis excluding events within the first 2 years of follow-up was not feasible given that 97.9% of events occurred at the 7-year wave.
5. Conclusions
In this CHARLS-based prospective cohort of Chinese adults with CKM stages 0–3, a higher MetS score was independently associated with elevated CVD risk in a graded, dose-response manner. The association persisted after adjustment for CKM stage and was robust across sensitivity analyses, though the incremental discriminative value beyond CKM staging was modest. The MetS score—a simple, clinically accessible metric—may serve as a complementary marker of cumulative metabolic burden within the CKM framework. Future studies with adjudicated outcomes, accurate event timing, repeated metabolic measurements, and formal incremental prediction analyses are needed to establish the potential clinical relevance of MetS scoring in CKM-based risk stratification.
CRediT authorship contribution statement
Li-Xin Cao: Writing – original draft. Dan Luo: Writing – review & editing. Shuang Liu: Investigation. Jun Qi: Supervision.
Funding
This study received no specific funding.
Conflicts of interest
The authors declare no conflicts of interest.
Footnotes
Supplementary data to this article can be found online at https://doi.org/10.1016/j.metop.2026.100498.
Appendix A. Supplementary data
The following is the Supplementary data to this article:
References
- 1.World Health Organization . WHO Fact Sheet; 2024. Cardiovascular diseases (CVDs) [Google Scholar]
- 2.Ndumele C.E., Rangaswami J., Chow S.L., et al. Cardiovascular-kidney-metabolic health: a presidential advisory from the American heart association. Circulation. 2023;148(20):1606–1635. doi: 10.1161/CIR.0000000000001184. [DOI] [PubMed] [Google Scholar]
- 3.Marassi M., Fadini G.P. The cardio-renal-metabolic connection: a review of the evidence. Cardiovasc Diabetol. 2023;22(1):195. doi: 10.1186/s12933-023-01937-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Ndumele C.E., Neeland I.J., Tuttle K.R., et al. A synopsis of the evidence for the science and clinical management of cardiovascular-kidney-metabolic (CKM) syndrome. Circulation. 2023;148(20):1636–1664. doi: 10.1161/CIR.0000000000001186. [DOI] [PubMed] [Google Scholar]
- 5.Rangaswami J., Bhalla V., de Boer I.H., et al. Cardiorenal protection with the newer antidiabetic agents. Circulation. 2020;142(17):e265–e286. doi: 10.1161/CIR.0000000000000920. [DOI] [PubMed] [Google Scholar]
- 6.Li W., Shen C., Kong W., et al. Association between the triglyceride glucose-body mass index and future cardiovascular disease risk in a population with CKM syndrome stage 0-3. Cardiovasc Diabetol. 2024;23(1):292. doi: 10.1186/s12933-024-02352-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Liu Y., Li W., Zhou H., et al. Association between atherogenic index of plasma and new-onset stroke in a population with CKM stages 0-3: insights from CHARLS. Cardiovasc Diabetol. 2025;24(1):168. doi: 10.1186/s12933-025-02732-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Alberti K.G., Eckel R.H., Grundy S.M., et al. Harmonizing the metabolic syndrome. Circulation. 2009;120(16):1640–1645. doi: 10.1161/CIRCULATIONAHA.109.192644. [DOI] [PubMed] [Google Scholar]
- 9.Mottillo S., Filion K.B., Genest J., et al. The metabolic syndrome and cardiovascular risk: a systematic review and meta-analysis. J Am Coll Cardiol. 2010;56(14):1113–1132. doi: 10.1016/j.jacc.2010.05.034. [DOI] [PubMed] [Google Scholar]
- 10.Klein B.E., Klein R., Lee K.E. Components of the metabolic syndrome and risk of cardiovascular disease and diabetes in beaver dam. Diabetes Care. 2002;25(10):1790–1794. doi: 10.2337/diacare.25.10.1790. [DOI] [PubMed] [Google Scholar]
- 11.Gami A.S., Witt B.J., Howard D.E., et al. Metabolic syndrome and risk of incident cardiovascular events and death. J Am Coll Cardiol. 2007;49(4):403–414. doi: 10.1016/j.jacc.2006.09.032. [DOI] [PubMed] [Google Scholar]
- 12.Galassi A., Reynolds K., He J. Metabolic syndrome and risk of cardiovascular disease: a meta-analysis. Am J Med. 2006;119(10):812–819. doi: 10.1016/j.amjmed.2006.02.031. [DOI] [PubMed] [Google Scholar]
- 13.Lakka H.M., Laaksonen D.E., Lakka T.A., et al. The metabolic syndrome and total and cardiovascular disease mortality in middle-aged men. JAMA. 2002;288(21):2709–2716. doi: 10.1001/jama.288.21.2709. [DOI] [PubMed] [Google Scholar]
- 14.Ford E.S. Risks for all-cause mortality, cardiovascular disease, and diabetes associated with the metabolic syndrome. Diabetes Care. 2005;28(7):1769–1778. doi: 10.2337/diacare.28.7.1769. [DOI] [PubMed] [Google Scholar]
- 15.Zhao Y., Hu Y., Smith J.P., Strauss J., Yang G. Cohort profile: the China health and retirement longitudinal study (CHARLS) Int J Epidemiol. 2014;43(1):61–68. doi: 10.1093/ije/dys203. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Tan C.E., Ma S., Wai D., Chew S.K., Tai E.S. Can we apply the NCEP ATP III definition of the metabolic syndrome to Asians? Diabetes Care. 2004;27(5):1182–1186. doi: 10.2337/diacare.27.5.1182. [DOI] [PubMed] [Google Scholar]
- 17.Ma Y.C., Zuo L., Chen J.H., et al. Modified glomerular filtration rate estimating equation for Chinese patients with chronic kidney disease. J Am Soc Nephrol. 2006;17(10):2937–2944. doi: 10.1681/ASN.2006040368. [DOI] [PubMed] [Google Scholar]
- 18.Isomaa B., Almgren P., Tuomi T., et al. Cardiovascular morbidity and mortality associated with the metabolic syndrome. Diabetes Care. 2001;24(4):683–689. doi: 10.2337/diacare.24.4.683. [DOI] [PubMed] [Google Scholar]
- 19.Bonora E., Kiechl S., Willeit J., et al. Carotid atherosclerosis and coronary heart disease in the metabolic syndrome. Diabetes Care. 2003;26(4):1251–1257. doi: 10.2337/diacare.26.4.1251. [DOI] [PubMed] [Google Scholar]
- 20.Navaneethan S.D., Schold J.D., Kirwan J.P., et al. Metabolic syndrome, ESRD, and death in CKD. Clin J Am Soc Nephrol. 2013;8(6):945–952. doi: 10.2215/CJN.09870912. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Sattar N., Gaw A., Scherbakova O., et al. Metabolic syndrome with and without C-reactive protein as a predictor of coronary heart disease and diabetes. Circulation. 2003;108(4):414–419. doi: 10.1161/01.CIR.0000080897.52664.94. [DOI] [PubMed] [Google Scholar]
- 22.Malik S., Wong N.D., Franklin S.S., et al. Impact of the metabolic syndrome on mortality from coronary heart disease, cardiovascular disease, and all causes in United States adults. Circulation. 2004;110(10):1245–1250. doi: 10.1161/01.CIR.0000140677.20606.0E. [DOI] [PubMed] [Google Scholar]
- 23.McNeill A.M., Rosamond W.D., Girman C.J., et al. The metabolic syndrome and 11-year risk of incident cardiovascular disease in the ARIC study. Diabetes Care. 2005;28(2):385–390. doi: 10.2337/diacare.28.2.385. [DOI] [PubMed] [Google Scholar]
- 24.Rodriguez-Colon S.M., Mo J., Duan Y., et al. Metabolic syndrome clusters and the risk of incident stroke: the ARIC study. Stroke. 2009;40(1):200–205. doi: 10.1161/STROKEAHA.108.523035. [DOI] [PubMed] [Google Scholar]
- 25.Boden-Albala B., Sacco R.L., Lee H.S., et al. Metabolic syndrome and ischemic stroke risk: Northern manhattan study. Stroke. 2008;39(1):30–35. doi: 10.1161/STROKEAHA.107.496588. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Grundy S.M. Metabolic syndrome: a multiplex cardiovascular risk factor. J Clin Endocrinol Metab. 2007;92(2):399–404. doi: 10.1210/jc.2006-0513. [DOI] [PubMed] [Google Scholar]
- 27.Hotamisligil G.S. Inflammation, metaflammation and immunometabolic disorders. Nature. 2017;542(7640):177–185. doi: 10.1038/nature21363. [DOI] [PubMed] [Google Scholar]
- 28.Brownlee M. Biochemistry and molecular cell biology of diabetic complications. Nature. 2001;414(6865):813–820. doi: 10.1038/414813a. [DOI] [PubMed] [Google Scholar]
- 29.Berneis K.K., Krauss R.M. Metabolic origins and clinical significance of LDL heterogeneity. J Lipid Res. 2002;43(9):1363–1379. doi: 10.1194/jlr.r200004-jlr200. [DOI] [PubMed] [Google Scholar]
- 30.Chobanian A.V., Alexander R.W. Exacerbation of atherosclerosis by hypertension. Arch Intern Med. 1996;156(17):1952–1956. [PubMed] [Google Scholar]
- 31.Gansevoort R.T., Correa-Rotter R., Hemmelgarn B.R., et al. Chronic kidney disease and cardiovascular risk. Lancet. 2013;382(9889):339–352. doi: 10.1016/S0140-6736(13)60595-4. [DOI] [PubMed] [Google Scholar]
- 32.Feigin V.L., Brainin M., Norrving B., et al. World stroke organization (WSO): global stroke fact sheet 2022. Int J Stroke. 2022;17(1):18–29. doi: 10.1177/17474930211065917. [DOI] [PubMed] [Google Scholar]
- 33.Cho J., Kim T.H., Son Y., et al. Cardiovascular risk across stages of cardiovascular-kidney-metabolic syndrome in South Korea, 2002-2022: a population-based cohort study. Medicine (Baltim) 2025;104(51) doi: 10.1097/MD.0000000000045960. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Zha B., Cai A., Yu H., Wang Z. The impact of cardiovascular-kidney-metabolic syndrome staging on depression risk: a national study. Heart Lung Circ. 2025;34(10):1131–1138. doi: 10.1016/j.hlc.2025.06.1017. [DOI] [PubMed] [Google Scholar]
- 35.Ma C., Zhao D.F., Zhai L., Huo R.R., Liu Y.H. Role of depressive symptoms in accelerating cardiovascular diseases progression in cardiovascular-kidney-metabolic syndrome stages 0-3: a nationwide prospective cohort study. Gen Hosp Psychiatry. 2025;96:183–192. doi: 10.1016/j.genhosppsych.2025.07.013. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.