This retrospective cohort study used data from the Korean National Health Insurance Database (NHID), a comprehensive resource that includes the medical claims, demographic data, and health examination records of nearly the entire Korean population. The NHID integrates data from biennial health-screening programs to promote the early detection and management of chronic diseases. These health checkups include measurements of dlifestyle behaviors, anthropometric parameters, and laboratory markers, providing a robust dataset for epidemiological research. This dataset has been described in prior publications [10–12].

This study included 1,500,959 adults who participated in the National Health Screening Program between 2009 and 2012. A total of 3,046 individuals were excluded, including participants aged 90 years or older (n = 374) and participants with missing values for any of the following variables: blood pressure (n = 35), smoking status (n = 166), body mass index (BMI; n = 67), waist circumference (n = 68), fasting glucose (n = 47), lipid profile (n = 1,176), or estimated glomerular filtration rate (eGFR; n = 1,286). The final cohort comprised of 1,497,913 participants. The participants were categorized into four stages: stage 0 or 1 (n = 495,261); stage 2 (n = 862,009); stage 3 (n = 94,864); and stage 4 (n = 45,779). Within each CKM syndrome stage, the participants were divided into three groups based on the TyG index tertiles: Group 1 (TyG index < 8.27), Group 2 (TyG index 8.27–8.81), and Group 3 (TyG index > 8.81). Outcomes were analyzed according to these classifications (Fig. 1). This study was approved by the Institutional Review Board (GURI 2024-12-021), and the requirement for informed consent was waived due to the anonymized and de-identified nature of the NHID dataset. All analyses adhered to relevant ethical guidelines, and the study complied with the tenets of the Declaration of Helsinki.

The key variables collected were demographic factors (age and sex), lifestyle behaviors (smoking status, alcohol consumption, and physical activity), and socioeconomic status (household income categorized into quartiles). Anthropometric measurements included BMI and waist circumference. Laboratory markers included fasting glucose, total cholesterol, high-density lipoprotein (HDL) cholesterol, low-density lipoprotein cholesterol, triglycerides, and eGFR. Blood pressure (systolic and diastolic) was measured. Clinical history variables included prior diagnoses of hypertension, diabetes mellitus (DM), and dyslipidemia. Medication use was documented for antihypertensive, glucose-lowering, lipid-lowering, and antiplatelet drugs.

The TyG index was calculated using the following formula: TyG index = ln [fasting triglycerides (mg/dL) × fasting glucose (mg/dL)/2] [13]. The participants were classified into three tertiles based on their TyG indices, relative to the distribution of the entire study population: Group 1 (< 8.27), Group 2 (8.27–8.81), and Group 3 (> 8.81), with Group 1

serving as the reference for comparative analysis. In addition, the receiver operating characteristic (ROC) curve analysis for the composite primary outcome was performed in the total population and across the CKM syndrome stages to identify the optimal cutoff values for the TyG index using Youden’s index (Supplementary Fig. 1). These cutoffs were subsequently used to dichotomize the participants into high- and low-TyG groups in the supplementary analysis.

The participants were stratified into CKM syndrome stages (0–4) based on previously established criteria incorporating metabolic, cardiovascular, and renal parameters [1,2]. Stage 0 represented individuals without cardiometabolic risk factors or CKD. Stage 1 included those with overweight or dysfunctional adiposity, but no additional metabolic or renal dysfunction. Stage 2 included participants with metabolic risk factors, such as hypertriglyceridemia, hypertension, metabolic syndrome, or diabetes, as well as patients with CKD, who exhibited eGFRs ranging from 30–59 mL/min/1.73 m². Stage 3 included individuals with subclinical CVD, defined by either very high-risk CKD (eGFR < 30 mL/min/1.73 m²) or an elevated 10-year cardiovascular risk, as indicated by a Predicting Risk of Cardiovascular Disease EVENTs (PREVENT) score ≥ 20% [14]. The PREVENT score—a cardiovascular risk model recently developed by the American Heart Association—is derived and validated using data from over six million individuals across 46 U.S.-based datasets. In this study, we used a base model, which incorporated variables, such as age, non-HDL cholesterol levels, HDL cholesterol levels, systolic blood pressure, diabetes, smoking status, BMI, eGFR, and use of antihypertensive or statin medications. The detailed formula is available at: https://professional.heart.org/prevent [14]. Stage 4 represented individuals clinically diagnosed with CVD. A detailed description of the CKM staging is provided in Supplementary Table 1.

The primary outcome of the study was a composite endpoint comprising all-cause death, heart failure, stroke (both ischemic and hemorrhagic), and myocardial infarction during the follow-up period, which concluded on December 31, 2022. The average follow-up duration was 12.60 ± 1.50 years. Secondary outcomes included the individual components of the primary composite outcomes. Heart failure was identified based on hospitalization records using the International Classification of Diseases, Tenth Revision (ICD-10) codes I50, I42.0, I11.0, or I13.0–I13.2. Myocardial infarction was defined as hospitalization with coronary revascularization and a discharge diagnosis coded I21 or I22. Stroke was confirmed in hospitalized individuals using brain imaging and discharge diagnoses of ICD-10 codes I63–I64 for ischemic stroke and I60–I62 for hemorrhagic stroke. Supplementary Table 2 provides a detailed list of diagnostic and procedural definitions, including the ICD-10 codes used to classify the comorbidities, CKM syndrome stages, and clinical outcomes.

Baseline characteristics according to the TyG index tertiles were compared using the chi-square test for categorical variables and one-way analysis of variance (ANOVA) for continuous variables. The incidence rates of cardiovascular outcomes were calculated as the total number of events divided by the cumulative person-years of follow-up and expressed per 1,000 person-years. Kaplan–Meier survival curves were constructed for each CKM syndrome stage to compare event-free survival across TyG tertiles. Statistical differences between survival curves were tested using the log-rank test. Cox proportional hazards regression models were used to assess the association between the TyG index tertiles and outcomes within each CKM syndrome stage. Hazard ratios (HRs) and 95% confidence intervals (CIs) were adjusted for confounders including age, sex, smoking status, alcohol consumption, physical activity, household income, and medication use (antihypertensive, glucose-lowering, lipid-lowering, and antiplatelet drugs). Model 1 included adjustments for age and sex, whereas Model 2 included additional lifestyle and socioeconomic variables. Model 3 was further adjusted for medication use (antihypertensive, glucose-lowering, lipid-lowering, and antiplatelet drugs); eGFR category (≥ 90, 60–89, 30–59, 15–30, < 15); and dipstick proteinuria. To assess the continuous relationship between the TyG index and cardiovascular outcomes, restricted cubic spline regression models were applied using Model 3 adjustments across all CKM syndrome stages. All analyses were performed using complete case data, excluding participants with missing data. Statistical significance was set at a two-tailed p value < 0.05. All the analyses were performed using SAS (version 9.4; SAS Institute, Cary, NC, USA) and R (version 4.2.1; R Foundation for Statistical Computing).

Table 1 presents the baseline characteristics of the study population stratified by TyG index tertiles. Participants in the higher TyG tertiles were older, with Group 3 having the highest mean age (50.83 ± 13.17 years) and a greater proportion of men (66.9% vs. 53.8% vs. 38.9%, Group 3 vs. 2 vs. 1). A higher TyG index was associated with increased metabolic risk factors, including an elevated BMI; waist circumference; and fasting glucose, triglyceride, and total cholesterol levels. The prevalence of hypertension, DM, and dyslipidemia increased progressively across the TyG tertiles, with Group 3 exhibiting the highest prevalence of these conditions. Similarly, the use of antihypertensive, glucose-lowering, lipid-lowering, and antiplatelet medications was most frequent in Group 3 (p < 0.001). In addition, the mean PREVENT score increased progressively across TyG tertiles (3.10 ± 6.08 in Group 1, 5.42 ± 7.55 in Group 2, and 8.72 ± 9.59 in Group 3; p < 0.001). The distribution of CKM syndrome stages differed notably across TyG tertiles (p < 0.001). In Group 1, a higher proportion of participants were classified as stage 0 or 1, whereas Group 3 was predominantly composed of individuals with stage 2 or higher. The proportions of participants in stages 3 and 4 were significantly higher in Group 3 (10.1% and 3.8%, respectively) than in Group 1 (3.2% and 2.1%, respectively).

Figure 2 illustrates the Kaplan–Meier survival curves for the composite primary outcomes (all-cause death, heart failure, stroke, or myocardial infarction) and the secondary outcomes across the TyG index tertiles in the total population. Individuals in the highest TyG tertile (Group 3) exhibited a significantly higher cumulative incidence of the composite primary outcome than those in the lower tertiles, demonstrating a clear dose–response relationship (Fig. 2A, log-rank p < 0.001). A similar pattern was observed for the secondary outcomes: all-cause death (Fig. 2B), heart failure (Fig. 2C), stroke (Fig. 2D), and myocardial infarction (Fig. 2E). Overall, these trends were consistent with the primary outcome, showing a higher cumulative incidence in the highest TyG tertile with a dose–response relationship (all log-rank p < 0.001).

Table 2 shows the association between the TyG index tertiles and the composite primary outcome after adjusting for potential confounders in both the total population and across the CKM syndrome stages. In the fully adjusted model (Model 3), the individuals in Group 3 exhibited a significantly higher risk of the composite primary outcome than those in Group 1 (HR, 1.116; 95% CI, 1.101–1.131; p < 0.001), showing a dose-dependent relationship. Group 2 showed a modest increase in risk (HR, 1.035; 95% CI, 1.021–1.049; p < 0.001). When stratified by CKM syndrome stage (interaction p < 0.001), a stronger association was observed in the earlier stages, with the highest HRs observed in Group 3 at stage 0/1 (HR, 1.166; 95% CI, 1.036–1.312; p = 0.011) and stage 2 (HR, 1.090; 95% CI, 1.069–1.111, p < 0.001). The effect remained significant in stage 3, although it was attenuated compared to that of the earlier stages (HR, 1.055; 95% CI, 1.025–1.085; p < 0.001). However, in Stage 4, the association between Group 3 and the composite outcome was no longer statistically significant after adjustment (HR, 1.040; 95% CI, 0.999–1.083; p = 0.055). Group 2 followed a similar trend, with a moderate increase in risk across all CKM syndrome stages. However, only the increased risk for stage 2 was significant (HR, 1.023; 95% CI, 1.004–1.044; p = 0.019). In addition, when the participants were dichotomized using the optimal TyG cutoff values identified by the ROC curve analysis for the composite outcome in the total population and CKM stages (8.489 for the total population, 8.097 for Stage 0/1, 8.132 for Stage 2, 9.571 for Stage 3, and 8.509 for Stage 4), those with a high TyG index had a higher risk of the primary composite outcome than those with a low TyG index relative to the total population and all CKM stages. For secondary outcomes (Supplementary Tables 3–6), Group 3 showed significantly higher risks for all-cause death (HR, 1.061; 95% CI, 1.042–1.081; p < 0.001); heart failure (HR, 1.054; 95% CI, 1.032–1.078; p < 0.001); stroke (HR, 1.209; 95% CI, 1.179–1.239; p < 0.001); and myocardial infarction (HR, 1.736; 95% CI, 1.647–1.829; p < 0.001) in the total population. The relative risks of myocardial infarction and stroke were higher than those of all-cause death and heart failure. Across CKM stages, the association showed a similar pattern to that of the primary outcome, with stronger effects observed in the earlier stages than in the later stages.

Furthermore, Figure 3 presents the RCS curves illustrating the association between the TyG index and composite primary outcome based on the fully adjusted model across the total population (Fig. 3A) and CKM syndrome stages (Fig. 3B–E). The analysis demonstrated a nonlinear, dose-dependent relationship, with a notable increase in HRs at higher TyG index levels. Excluding Stage 0/1, in which the TyG index distribution was concentrated at lower values, earlier CKM syndrome stages exhibited a greater increase in HRs at higher TyG index levels than later stages.

To assess the potential effect of lipid-lowering therapy, we conducted a subgroup analysis stratified by baseline statin use. Although the association between a higher TyG index and clinical outcomes remained generally consistent across the statin user and non-user groups, the magnitude of the association appeared slightly more pronounced among non-users. The interaction p values were statistically significant for heart failure and myocardial infarction (Supplementary Table 7).