The National Health and Nutrition Examination Survey (NHANES) employs a dual methodology that combines structured interviews and physical examinations to evaluate the health and nutritional status of adults and children in the United States. The interviews collect demographic, socioeconomic, dietary, and health-related information, while the examinations provide objective clinical and laboratory data. The NHANES protocol is approved by the National Centre for Health Statistics (NCHS) Ethics Review Board, and all participants provide written informed consent. The survey is conducted in two-year cycles, and this study utilized data from seven consecutive cycles between 2005 and 2018, with mortality follow-up through December 2019. Further information about NHANES is available at https://wwwn.cdc.gov/Nchs/Nhanes/.

Initially, 51,199 participants from NHANES were considered. After excluding individuals younger than 20 years or those with missing CKM-related diagnostic data, incomplete ePWV information, insufficient mortality follow-up, or uncertain medical history, 14,372 participants were included in the final analysis. Uncertain medical history referred to incomplete, inconsistent, or unreliable information regarding CKM-related conditions—including cardiovascular disease, diabetes, chronic kidney disease, obesity, and hypertension—or missing/conflicting laboratory and examination data required for CKM classification. These participants were excluded to ensure diagnostic accuracy and analytic reliability. Missing data were evaluated for all covariates, and when the proportion of missing values was below 15%, multiple imputation by chained equations (MICE) was applied to preserve the largest possible sample size and minimize bias from incomplete data. The participant selection process is illustrated in Figure 1.

ePWV, a non-invasive measure derived from age and mean blood pressure, was used to assess arterial stiffness, an indicator of vascular ageing (9, 10). The ePWV was calculated from age and mean blood pressure (MBP) using the following equation: 9.587 − 0.402 × age + 4.560 × 10^-³ × age² − 2.621 × 10^-5 × age² × MBP + 3.176 × 10 ³ × age × MBP − 1.832 × 10^-² × MBP. MBP was calculated as diastolic blood pressure + 0.4 × (systolic blood pressure − diastolic blood pressure).

CKM was assessed using the American Heart Association (AHA) Presidential Advisory(1). Each CKM stage (0–4) was operationalized using National Health and Nutrition Examination Survey (NHANES) variables, following the AHA criteria that integrate metabolic risk factors, kidney function, and cardiovascular status. The detailed mapping of CKM stages, including variable definitions, diagnostic thresholds, and decision rules, is provided in Supplementary Table S1. These definitions were used to determine the outcome variable in all subsequent analyses. For missing data, we employed a complete-case analysis approach, excluding participants with missing key variables for CKM staging (e.g., eGFR, albuminuria, or CVD history). No imputation was performed, and “unknown” responses were treated as missing.

Identification of mortality status was achieved through the synchronization of NHANES datasets and records maintained by the National Death Index (NDI), accessible via the portal at https://www.cdc.gov/nchs/data-linkage/mortality-public.htm. From the NDI’s repository, participants’ survival or demise classification was ascertained accordingly. The International Classification of Diseases, Tenth Revision (ICD-10), was utilized to define the underlying causes of death. Cardiovascular mortality was classified explicitly by the NCHS as deaths attributed to heart disease, identified by ICD-10 codes I00-I09, I11, I13, and I20-I51 (11).

An external validation cohort was constructed using data from a retrospective study conducted in the Department of Cardiology at a tertiary hospital in southern China. A total of 3,157 patients were included after excluding individuals with missing information on CKM staging or ePWV. The cohort primarily consisted of middle-aged and older adults (mean age 47.98 ± 12.88 years), with 57.43% male participants. The prevalence of hypertension, diabetes, and chronic kidney disease was 45.62%, 28.83%, and 15.78%, respectively. CKM classification was defined using the same diagnostic criteria as applied in the NHANES cohort to ensure consistency. This independent dataset was used to validate the ePWV cutoff derived for Stage 4 CKM patients in the NHANES population. The same modelling framework and covariate adjustment strategy were adopted to ensure comparability between the two cohorts. For ethical considerations, all NHANES data are publicly available and de-identified, with informed consent obtained by the National Centre for Health Statistics (NCHS) at the time of data collection; no additional ethical approval was required for secondary analyses. The external validation cohort, derived from anonymized hospital records, was approved by the Institutional Review Board of The First People’s Hospital of Yulin (Approval No. YLSY-IRB-SR-2025114), which formally waived the requirement for individual informed consent in accordance with national and institutional regulations.

Demographic information, including age, sex, race, educational attainment, marital status, family poverty income ratio (PIR), body mass index (BMI), and tobacco and alcohol use, was collected using standardized questionnaires and a computer-assisted interview system. See the NHANES Laboratory/Medical Technologists Procedures Manual for more details (12). Diabetes mellitus (DM) encompasses either an individual’s self-disclosure of the condition, their reporting of antidiabetic medication consumption, or a fasting plasma glucose concentration of≥ 126 mg/dL. Cardiovascular diseases (CVD) in discourse include heart failure, coronary artery disease, episodes of angina pectoris, myocardial infarctions, and cerebrovascular accidents or strokes within their scope. According to the guidelines set by the Kidney Disease: Improving Global Outcomes (KDIGO), estimates of populations affected by chronic kidney diseases were undertaken. Hyperlipidemia, when defined within these parameters, targets individuals taking lipid-reducing pharmaceuticals or those registering LDL levels at or above 140 mg/dL; this consideration also extends to males exhibiting HDL < 40 mg/dL and females whose HDL < 50 mg/dL. Regarding hypertension, its delineation involves a combination of patient-asserted medical diagnoses, the declared use of antihypertensive drugs, and recordings of systolic pressure meeting or exceeding 140 mmHg and/or diastolic pressure ≥ 90 mmHg (12).

Baseline information was expressed as weighted means (standard error) for continuous variables and weighted percentages for categorical variables. Intergroup differences were compared using weighted variance and chi-square tests. The relationships between ePWV and CKM were analyzed using weighted multivariable logistic regression, with covariates progressively incorporated to minimize confounding. The quartile-based grouping enables a nonparametric assessment of the dose–response relationship between arterial stiffness and CKM risk, thereby avoiding the assumption of linearity inherent in continuous models. The present study utilized univariable and multivariable weighted Cox proportional hazards models to assess the relationships between ePWV and all-cause and CKM mortality in patients with CKM. The outcomes of this analysis included hazard ratios (HRs) and 95% confidence intervals (CIs). Restricted cubic spline (RCS) regression was utilized to examine the potential for a dose-response relationship between ePWV and CKM, with the nonlinearity of the relationship being assessed via likelihood ratio tests. The investigation involved stratified analyses, which were undertaken to examine the consistency of the relationship between ePWV and CKM across diverse demographic subgroups. Subgroups were analyzed in order to explore the association between ePWV and CVD prevalence. This analysis was based on various demographic characteristics. The age of subjects, smoking status, gender, and CKM syndrome stage were considered. The effects of metabolic indicators were analyzed to see how they mediate the effects of age, smoking, and CKM. The present study, therefore, sought to evaluate the mediating effect of the Triglyceride-Glucose Index (TyG) on the relationships of age and smoking with CKM. The mediating effect was quantified by deriving a percentage, calculated as the indirect effect divided by the total effect. The hypothesis concerning the significance of the mediating effect was tested using a bootstrap sample (n = 1000) (13). All analyses accounted for the complex, multistage probability sampling design of NHANES. Sampling weights, strata, and primary sampling units (PSUs) were incorporated in accordance with NHANES analytic guidelines. Because this analysis included laboratory and examination variables collected through the Mobile Examination Centre (MEC), the MEC examination weight (WTMEC2YR) was selected based on the principle of using the smallest applicable subsample weight. When combining seven 2-year survey cycles (2005–2018), the 2-year MEC weights (WTMEC2YR) were divided by seven to construct an appropriate 14-year combined weight representing the U.S. civilian, noninstitutionalized population. Variance estimates were computed using the Taylor series linearization method, specifying SDMVSTRA as the strata variable and SDMVPSU as the primary sampling unit. All analyses were performed using the survey procedures in R (package survey) to ensure nationally representative estimates.

A total of 14,372 participants were included in the final analysis, representing approximately 88 million U.S. adults after weighting and nesting. Participants were stratified based on the presence or absence of CKM. Among those with CKM, 43.31% were female and 35.37% were young adults aged 20–45 years (Table 1). In the smoking subgroup, the prevalence of CKM was nearly twice as high as that of participants without depressive symptoms. Baseline differences between the MASLD and control groups were assessed using t-tests or chi-square tests, revealing significant differences in age, race, education level, smoking status, and other variables (P < 0.01). Additionally, ePWV scores were significantly higher in the CKM group compared to the control group (P < 0.01).

Subgroup analyses revealed a consistent positive association between ePWV and CKM across all subgroups, including sex, age, race, poverty-income ratio (PIR), and smoking status. Significant interactions were observed for age, race, and smoking (P for interaction < 0.01), as shown in Figure 2. Table 2 summarizes the association between ePWV and CKM. After full adjustment for covariates (Model 3), the association remained significant but attenuated, with an odds ratio (OR) of 1.31 (95% CI: 1.27–1.37, P < 0.01). Participants in the highest ePWV quartile (Q4) had a substantially higher risk of CKM compared to those in the lowest quartile (Q1), with an OR of 4.68 (95% CI: 3.77–5.83, P < 0.01). This positive trend was consistent across quartiles (P for trend < 0.01). Restricted cubic spline (RCS) analysis revealed a significant nonlinear positive association between ePWV and CKM risk (P for overall < 0.01; P for nonlinearity < 0.01). As shown in Figure 3, CKM risk demonstrated a generally positive association with increasing ePWV values, indicating that higher arterial stiffness was related to greater cardiometabolic risk. The regression model identified an inflection point around 14.7 m/s; however, this estimate is likely influenced by the limited number of participants at the upper ePWV range and does not represent a clinically meaningful threshold.

Logistic regression based on the fully adjusted model was used to assess the association between ePWV and different stages of CKM. The results demonstrated a stepwise increase in the odds ratios (ORs) with advancing CKM stages: 1.00 for Stage 1, 1.44 (95% CI: 1.41–1.47) for Stage 2, 1.61 (95% CI: 1.55–1.67) for Stage 3 (preclinical CVD), and 1.85 (95% CI: 1.79–1.91) for Stage 4 (clinical CVD), all with P < 0.01. More details can be seen in Table 3. Sensitivity analyses further supported this trend. The AUC values progressively increased from 0.52 in Stage 1 to 0.72, 0.84, and 0.86 in Stages 2, 3, and 4, respectively, indicating an improved discriminative ability of ePWV with increasing disease severity. Similar increasing trends were observed for sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV), suggesting that ePWV is a robust marker across CKM stages, particularly for detecting advanced CKM with cardiovascular involvement. More details can be seen in Table 4.

A retrospective study of a Southern Chinese population was conducted to construct an external validation cohort. After excluding cases with missing CKM staging and ePWV data, a total of 3,157 patients with clinical cardiovascular diseases (CVD) in CKM stage IV were included. This independent dataset was used to assess the applicability of the regression analysis results from the NHANES cohort and the estimated ePWV threshold for stage IV. Using generalized additive models (GAMs), we estimated the ePWV thresholds corresponding to different stages of CKM based on data from NHANES stages 1 to 4. After adjusting for age, gender, race/ethnicity, poverty income ratio (PIR), education, and smoking, the thresholds increased steadily with disease progression: 7.27 for Stage 1, 7.80 for Stage 2, 8.35 for Stage 3, and 8.81 for Stage 4 (all P < 0.01). To examine whether the Stage 4 threshold could be applied beyond the modeling dataset, we tested it in an independent external cohort. The threshold identified in the validation cohort was 8.52, slightly lower than the estimate from the NHANES-derived model. Despite this difference, the threshold remained statistically significant (P < 0.01), and the overall trend was consistent. More details can be seen in Table 5.

As shown in Figure 4, Kaplan–Meier survival analysis demonstrated that patients with higher ePWV levels had significantly lower survival probabilities compared with those with lower ePWV (median survival: 66 vs. 89 months; log-rank P < 0.01). The restricted mean survival time (RMST) was also shorter in the high-ePWV group (79.41 vs. 108.71 months), indicating that elevated ePWV was associated with poorer prognosis among individuals with CKM. To further examine this association, Cox proportional hazards regression models were applied to estimate the relationship between ePWV and mortality risk. As summarized in Table 6, higher ePWV was consistently associated with increased risks of both cardiovascular (CVD) and all-cause mortality. In the unadjusted model, each unit increase in ePWV was associated with a 43% higher risk of CVD mortality (HR = 1.43, 95% CI 1.38–1.48, P < 0.01) and a 25% higher risk of all-cause mortality (HR = 1.25, 95% CI 1.43–1.49, P < 0.01). After adjusting for demographic and clinical covariates, these associations remained robust. In the fully adjusted model, ePWV remained a significant predictor of CVD mortality (HR = 1.15, 95% CI 1.12–1.25, P < 0.01) and all-cause mortality (HR = 1.10, 95% CI 1.27–1.37, P < 0.01). A graded association was observed across ePWV quartiles, with progressively higher mortality risks in higher categories (P for trend < 0.01). Compared with participants in the lowest ePWV quartile, those in the highest quartile had a 5.91-fold higher risk of CVD mortality (95% CI 5.32–6.75) and a 3.66-fold higher risk of all-cause mortality (95% CI 2.55–3.28) in the fully adjusted model.

Mediation analysis was conducted to explore the potential role of the triglyceride–glucose (TyG) index in the association between estimated pulse wave velocity (ePWV) and cardiovascular–kidney–metabolic syndrome (CKM). As illustrated in Figure 5, TyG significantly mediated this relationship, with an indirect effect (IE) of 0.0103 (95% CI: 0.0069–0.0135, P < 0.01) and a direct effect (DE) of 0.0307 (95% CI: 0.0248–0.0358, P < 0.01). The proportion of the total effect mediated by TyG was 25.23%, indicating a substantial metabolic contribution to the observed link between arterial stiffness and CKM.