This study retrospectively analyzed a nationwide cohort of patients who underwent health checkups between 2009 and 2012, as provided by the National Health Insurance Service (NHIS) of South Korea. The NHIS database includes individual health checkup results, along with claims data containing diagnostic, procedural, and prescription codes. As the NHIS serves as the sole public insurer in South Korea, its database is representative of the Korean population. The health checkup database comprises demographic data, including body measurements, blood and urine test results, and survey results on health-related behaviors.

This study identified 6,891,401 individuals aged 20‐39 years who underwent a national health checkup between 2009 and 2012. We excluded 330,334 individuals with insufficient data to calculate the TyG index; 2412 individuals with ESKD identified using rare incurable disease registry codes for hemodialysis (V001), peritoneal dialysis (V003), or kidney transplantation (V005); and 2732 individuals with insurance claims related to ESKD or who died within 1 year after the index date. The remaining 6,555,863 eligible individuals were included and followed up until December 2021. Regarding the time frame, the start of follow-up (time zero) was defined as 1 year after the index date (lag period). Follow-up was stopped at death, censorship (eg, immigration), or ESKD development. The mean follow-up duration was 10.34 (SD 1.18) years. The participants were divided into 4 groups based on their TyG index, with quartiles 1 and 4 representing the lowest and the highest values, respectively. The mean (SD) values of the TyG index for each quartile were as follows: quartile 1: 7.68 (SD 0.29), quartile 2: 8.16 (SD 0.27), quartile 3: 8.52 (SD 0.32), and quartile 4: 9.15 (SD 0.51).

All participants completed a questionnaire detailing their medical history and lifestyle habits, including smoking, alcohol consumption, and exercise. Smoking status was categorized as nonsmoker, former smoker, or current smoker. Alcohol consumption was classified as nondrinker, mild drinker (<30 g/day), or heavy drinker (≥30 g/day). Regular exercise was defined as vigorous-intensity exercise performed 3 or more times per week or moderate-intensity exercise performed 5 or more times per week. Routine physical examinations included body weight, height, waist circumference, and blood pressure. BMI was calculated as body weight (kg) divided by height squared (m²). A low income was defined as being in the lowest quartile of the NHIS population or receiving government-funded medical aid. Blood samples were collected after overnight fasting to assess the serum levels of glucose, creatinine, total cholesterol, triglycerides, high-density lipoprotein-cholesterol (HDL-C), and low-density lipoprotein-cholesterol. The estimated glomerular filtration rate (eGFR) was calculated using the Modification of Diet in Renal Disease equation [17].

The TyG index was calculated using the formula log (fasting triglyceride [mg/dL] × fasting blood glucose [mg/dL]/2) [13,18]. Baseline comorbidities were defined as follows: hypertension (ICD-10 [International Statistical Classification of Diseases and Related Health Problems, 10th Revision] codes I10-I13 or I15 and treatment with antihypertensive medications, systolic blood pressure [SBP] ≥ 140 mmHg, or diastolic blood pressure [DBP] ≥ 90 mmHg), diabetes (ICD-10 codes E11-E14 and antidiabetic treatment or fasting glucose level ≥ 126 mg/dL), dyslipidemia (ICD-10 code E78 with lipid-lowering agents or serum total cholesterol ≥240 mg/dL), and CKD (eGFR<60 mL/min/1.73 m²). Antihypertensive medications included angiotensin type II receptor blockers, angiotensin-converting enzyme inhibitors, calcium channel blockers, beta-blockers, diuretics, or alpha-blockers. Antidiabetic medications included metformin, meglitinide, sulfonylurea, dipeptidyl peptidase IV inhibitor, alpha-glucosidase inhibitor, thiazolidines, or insulin. Lipid-lowering agents (dyslipidemia medication) included statins, ezetimibe, or fibrates.

In this study, obesity was defined as a BMI ≥25 kg/m², consistent with Asian-specific criteria [19]. Metabolic syndrome (MetS) was diagnosed based on the revised National Cholesterol Education Program Adult Treatment Panel III criteria [20,21]. MetS was diagnosed when three or more of the following five criteria were met: (1) elevated blood pressure (SBP ≥130 mmHg, DBP ≥85 mmHg, use of antihypertensive medications at baseline, and/or a history of hypertension), (2) elevated triglycerides (fasting triglycerides  ≥ 150 mg/dL or use of dyslipidemia medications at baseline), (3) reduced HDL-C (<40 mg/dL for men and <50 mg/dL for women), (4) elevated fasting plasma glucose level (≥100 mg/dL or use of medications for elevated glucose levels), and (5) abdominal obesity (waist circumference ≥90 cm for men or ≥85 cm for women).

The primary outcome of this study was newly diagnosed ESKD, defined as the initiation of kidney replacement therapy and/or kidney transplantation. Kidney replacement therapy or kidney transplantation was defined based on the rare incurable disease registry codes for hemodialysis (V001), peritoneal dialysis (V003), or kidney transplantation (V005).

Categorical variables are presented as frequencies and percentages, whereas continuous variables are expressed as mean (SD). For continuous variables not following a normal distribution, log transformation was applied, and the results were presented as geometric means with 95% CIs. Baseline characteristics across the TyG index quartiles were compared using one-way ANOVA for continuous variables and the chi-square test for categorical variables. The follow-up duration was calculated for each TyG index quartile group. The incidence rates of ESKD were determined for the TyG index, fasting glucose, and triglyceride quartiles over the total follow-up period for the study outcome. Incidence rates were expressed as events per 1000 person-years. Kaplan-Meier curves were generated to estimate incidence trends, and the log-rank test was used to compare groups. All outcomes were analyzed using Cox proportional hazards regression analysis while controlling for baseline covariates. Sensitivity analyses were performed to reduce potential bias and demonstrate the consistency of our findings. Incidence rates of ESKD were calculated for deciles of the TyG index, fasting glucose, and triglycerides. Subgroup analyses were performed to explore whether the association between the TyG index and ESKD differed by age, sex, obesity, MetS, diabetes, diabetes medication, dyslipidemia, dyslipidemia medication, and CKD using Cox proportional hazard regression analysis adjusted for age, sex, BMI, smoking, drinking, regular exercise, low income, diabetes medication, dyslipidemia medication, hypertension, eGFR, and CKD. Additionally, 2-year follow-up values of the TyG index, fasting glucose, and triglyceride were used to analyze the incidence rates of ESKD and the association between the TyG index quartiles and ESKD. The incidence rates and associations were evaluated both in terms of the 2-year follow-up values and the time-averaged values of baseline and 2-year follow-up. The risks of ESKD associated with the TyG index groups were expressed as hazard ratios (HRs) with 95% CIs, using the first quartile or decile group as the reference. A two-tailed P<.05 was considered statistically significant. All analyses were performed using SAS (version 9.4; SAS Institute) and R (version 3.4.1; R Foundation for Statistical Computing).

This study adhered to the Declaration of Helsinki (2013 revision) and was approved by the Institutional Review Board of CHA Bundang Medical Center (CHAMC 2024-04-034). The requirement for written informed consent was waived by the Institutional Review Board because of the retrospective nature of the study.

Baseline demographic and clinical characteristics were compared across the TyG index quartiles (Table 1). An increase in the TyG index quartile was associated with significant increases in age, BMI, waist circumference, SBP, DBP, and levels of fasting plasma glucose, total cholesterol, and triglycerides, whereas HDL-C and eGFR levels decreased. Similarly, higher TyG index quartiles were associated with significantly greater proportions of individuals with high BMI, current smoking status, heavy alcohol consumption, and use of diabetes or dyslipidemia medication. The proportion of individuals with low income varied across the TyG index quartiles, with lower prevalence in the middle quartiles. Additionally, the prevalence of hypertension, diabetes, and dyslipidemia increased across quartiles, whereas the proportion of individuals engaging in regular exercise decreased. Elevated low-density lipoprotein-cholesterol levels and a higher prevalence of CKD were also observed in the higher TyG index quartiles.

The incidence rates, absolute incidence difference, 5- and 10-year cumulative incidence, and HRs of ESKD were analyzed across quartiles of the TyG index, fasting glucose, and triglycerides (Table 2). The incidence of ESKD increased progressively with higher TyG index quartiles (Figure 1). In both the unadjusted and adjusted models, the HRs of ESKD also increased significantly with increasing TyG index quartiles (all P<.001). In the fully adjusted Model 3, individuals in the highest TyG index quartile had an HR of 3.438 (95% CI 3.096‐3.818) compared with those in the lowest quartile.

Regarding fasting glucose, ESKD incidence rates and HRs were significantly higher in the highest quartile than in the lower quartiles (P<.001). The HR for ESKD in the highest fasting glucose quartile was 1.393 (95% CI 1.284‐1.511) compared with that in the lowest quartile in fully adjusted Model 3. However, the second (HR 0.836, 95% CI 0.761‐0.918) and third (HR 0.891, 95% CI 0.810‐0.979) quartiles showed significantly lower HRs compared to the lowest quartile group in adjusted Model 3.

For triglycerides, ESKD incidence rates and HRs showed patterns similar to those of the TyG index. The incidence rates increased progressively with the triglyceride quartiles. HRs for ESKD also increased significantly with increasing the triglyceride quartiles (P<.001). In the fully adjusted Model 3, the HR of the highest triglyceride quartile was 2.675 (95% CI 2.423‐2.953) compared to that of the lowest quartile. The second (HR 1.463, 95% CI 1.314‐1.628) and third (HR 1.693, 95% CI 1.527‐1.876) quartiles also showed significantly higher HRs than the lowest quartile in adjusted Model 3.

ESKD incidence rates and HRs were assessed across deciles of the TyG index, fasting glucose, and triglyceride (Table 3). The incidence rates of ESKD increased progressively with increasing TyG index deciles. Similarly, HRs for ESKD showed a significant upward trend as TyG index deciles increased in both the unadjusted and adjusted models (all P<.001). In the fully adjusted Model 3, individuals in the highest TyG index decile had an HR of 5.944 (95% CI 5.022‐7.035) compared to those in the lowest decile.

Regarding fasting glucose, the incidence rates and HRs of ESKD were significantly higher in the highest decile than in the lower deciles (P<.001). In the fully adjusted Model 3, the HR for ESKD in the highest fasting glucose decile was 2.029 (95% CI 1.805‐2.282) compared to the lowest decile. However, the second through ninth deciles did not show statistically significant HRs compared to the lowest decile in the same model.

For triglycerides, ESKD incidence rates and HRs showed similar patterns to those of the TyG index. The incidence rates of ESKD increased progressively with increasing triglyceride deciles. Similarly, HRs for ESKD gradually increased with higher triglyceride deciles in both the unadjusted and adjusted models (all P<.001). In the fully adjusted Model 3, the HR for ESKD in the highest triglyceride decile was 4.042 (95% CI 3.433‐4.759) compared to that in the lowest triglyceride decile.

The incidence rates and adjusted hazard ratios (aHRs) of ESKD were evaluated across TyG index quartiles within various subgroups (Table 4). Interaction tests revealed that age, sex, obesity, MetS, diabetes, use of medications for diabetes or dyslipidemia, dyslipidemia, low income, eGFR, and CKD significantly influenced the association between the TyG index and ESKD (all P for interaction<.0001).

In subgroups stratified by age (20‐29 years and 30‐39 years), sex (male and female), obesity status (with or without), and CKD status (with or without), both the incidence rates and aHRs for ESKD significantly increased with higher TyG index quartiles. For participants without MetS, diabetes, or dyslipidemia and those not receiving medications for diabetes or dyslipidemia, the incidence rates and aHRs of ESKD also rose consistently across increasing TyG index quartiles.

In participants with dyslipidemia, the aHR for ESKD was significantly higher in the highest TyG index quartile than in the lowest quartile; however, the aHRs for the second and third quartiles were not significant. Among participants with diabetes, the aHRs were lower in the second and third quartiles, and the highest quartile did not show a significant difference compared with the first quartile. For participants with MetS and those receiving medications for diabetes or dyslipidemia, the aHRs for ESKD were not significantly different between the TyG index quartiles.

Sensitivity analysis was performed using TyG index quartiles at 2-year follow-up and time-averaged values from baseline and 2-year follow-up (Table 5). The incidence rates and aHRs of ESKD across TyG index quartiles at 2-year follow-up and for time-averaged values showed findings consistent with those of the baseline TyG index quartiles.

The incidence of ESKD increased progressively with higher TyG index quartiles at 2-year follow-up. In both the unadjusted and adjusted models, the HRs of ESKD also increased significantly with increasing TyG index quartiles (all P<.001). In the fully adjusted Model 3, individuals in the highest TyG index quartile had an HR of 2.860 (95% CI 2.466‐3.317) compared with those in the lowest quartile.

Regarding fasting glucose at 2-year follow-up, ESKD incidence rates and HRs were significantly higher in the highest quartile than in the lower quartiles (P<.001). The HR for ESKD in the highest fasting glucose quartile was 1.306 (95% CI 1.156‐1.475) compared to that in the lowest quartile in the fully adjusted Model 3. However, the second (HR 0.845, 95% CI 0.737‐0.973) and third (HR 0.838, 95% CI 0.729‐0.963) quartiles showed significantly lower HRs compared with the lowest quartile group in adjusted Model 3.

For triglycerides at 2-year follow-up, ESKD incidence rates and HRs showed patterns similar to those of the TyG index. The incidence rates increased progressively with the triglyceride quartiles. HRs for ESKD also increased significantly with increasing the triglyceride quartiles (P<.001). In the fully adjusted Model 3, the HR of the highest triglyceride quartile was 2.174 (95% CI 1.884‐2.508) compared to that of the lowest quartile. The second (HR 1.253, 95% CI 1.074‐1.461) and third (HR 1.445, 95% CI 1.246‐1.677) quartiles also showed significantly higher HRs than the lowest quartile in adjusted Model 3.

The incidence of ESKD increased progressively with higher time-averaged TyG index quartiles. In both the unadjusted and adjusted models, the HRs of ESKD also increased significantly with increasing TyG index quartiles (all P<.001). In the fully adjusted Model 3, individuals in the highest TyG index quartile had an HR of 3.449 (95% CI 2.937‐4.050) compared with those in the lowest quartile.

Regarding time-averaged fasting glucose, ESKD incidence rates and HRs were significantly higher in the highest quartile than in the lower quartiles (P<.001). The HR for ESKD in the highest fasting glucose quartile was 1.565 (95% CI 1.389‐1.763) compared to that in the lowest quartile in fully adjusted Model 3. However, the second (HR 0.906, 95% CI 0.788‐1.042) and third (HR 0.770, 95% CI 0.669‐0.886) quartiles showed significantly lower HRs compared with the lowest quartile group in adjusted Model 3.

For time-averaged triglycerides, ESKD incidence rates and HRs showed patterns similar to those of the TyG index. The incidence rates increased progressively with higher triglyceride quartiles. HRs for ESKD also increased significantly with increasing the triglyceride quartiles (P<.001). In fully adjusted Model 3, the HR of the highest triglyceride quartile was 2.092 (95% CI 1.810‐2.417) compared to that of the lowest quartile. The second (HR 1.320, 95% CI 1.131‐1.542) and third (HR 1.475, 95% CI 1.268‐1.715) quartiles also showed significantly higher HRs than the lowest quartile in adjusted Model 3.

This observational, population-based cohort study found that a higher TyG index was significantly associated with an increased risk of incident ESKD in young adults. This association was independent of traditional CKD risk factors and remained robust across subgroups based on age, sex, obesity, dyslipidemia, and CKD. These findings suggest that the TyG index, a marker of insulin resistance, may be useful for identifying young adults at risk of developing ESKD, and early intervention targeting insulin resistance may mitigate the pathogenesis of ESKD in young adults.

CKD is a leading global cause of mortality, affecting over 10% of the general population owing to the rising risk factors such as hypertension, diabetes, and obesity [1]. CKD progression to ESKD, which requires kidney replacement therapy, represents a significant global health issue. According to the International Society of Nephrology’s 2023 Global Kidney Health Atlas, the median incidence of treated ESKD was 146 individuals per million population in 84 countries, while the median prevalence of treated ESKD was 823 per million population in 95 countries [22]. Although CKD and ESKD are more prevalent in older adults, the global incidence of CKD is increasing in adolescents and young adults [3], who experience 30 times higher mortality rates than their healthy peers [23]. They face accumulating physiological complications and socioeconomic burdens from early life, predisposing them to poor long-term outcomes [7,24]. Therefore, identifying CKD risk factors early and implementing prompt interventions are critical for reducing the disease burden among young adults.

In this study, young adults with a higher TyG index were older, had a lower eGFR, and exhibited poorer metabolic health profiles than those with a lower TyG index. Moreover, they had higher BMI, waist circumference, blood pressure, and fasting plasma glucose, and poorer lipid profiles. Additionally, individuals with a higher TyG index showed higher rates of obesity, smoking, and alcohol consumption, as well as lower rates of regular exercise. They also had a greater prevalence of hypertension, diabetes, and dyslipidemia and were more likely to be on medications for diabetes or dyslipidemia than those with a lower TyG index. These findings align with previous studies [15,25,26]. During follow-up, ESKD incidence rates increased progressively with increasing TyG index, and the risk of ESKD significantly increased across both quartiles and deciles, independent of various ESKD risk factors. The individual contributions of the 2 TyG index components (ie, fasting plasma glucose and triglyceride levels) were separately investigated by analyzing the incidence rates and HRs across quartiles and deciles. Although fasting plasma glucose showed no significant effect, triglyceride levels demonstrated a weak effect in the decile-based analysis, which is consistent with a previous report [15]. Additionally, the impact of a high TyG index on ESKD pathogenesis was robust in subgroup analyses based on age, sex, obesity, dyslipidemia, and CKD. Sensitivity analyses using the 2-year follow-up TyG index and the time-averaged TyG index also showed significant associations with ESKD development. These findings suggest that the TyG index serves as more than a simple sum of its components (fasting plasma glucose and triglyceride levels) and has the potential to be a marker for assessing ESKD risk in young adults.

Previous studies have shown that the TyG index mediates the relationship between BMI and ESKD risk in middle-aged adults [26] and in the general population [15]. Among adolescents and young adults, higher SBP, fasting plasma glucose levels, and BMI have been identified as important factors contributing to CKD risk [3]. In this study, an elevated TyG index was associated with incident ESKD in young adults with both normal and high BMI, suggesting that a high TyG index, reflective of insulin resistance, plays a role in ESKD development both in connection with BMI and independently. Insulin resistance is known to cause glomerular hyperfiltration, sodium retention, tubular dysfunction, kidney inflammation, and fibrosis [11], all of which may contribute to ESKD progression through pathways independent of BMI. Additionally, previous research has linked the TyG index to the progression of coronary artery calcification in patients with CKD [27], suggesting that insulin resistance may further exacerbate ESKD by promoting atherosclerosis and vascular calcification. Insulin resistance contributes to atherogenesis and atherosclerosis progression by inducing macrophage apoptosis and causing damage to endothelial and vascular smooth muscle cells [28,29]. Hyperinsulinemia also promotes oxidative stress, structural and functional vessel injury [30], and osteogenic differentiation resulting in vascular calcification [31]. These findings highlight the importance of managing metabolic factors, in addition to weight reduction, to mitigate the detrimental effects of insulin resistance on progressive kidney injury.

Unexpectedly, the subgroup analysis showed that the TyG index was not associated with the risk of ESKD in young adults with diabetes or MetS, or those on medications for diabetes or dyslipidemia. Recent research has linked the TyG index to an increased risk of ESKD in patients with diabetes [16]. The reason for the discrepancy in our study is unclear. We speculate that there are several possible explanations. First, the population in this study differed from that in previous research, as this analysis focused exclusively on young adults aged 20‐39 years. Similarly, a recent study analyzing the National Health and Nutrition Examination Survey found no significant association between the TyG index and low eGFR (<60 mL/min/1.73 m²) in individuals aged 20‐40 years [25]. Second, among young adults with diabetes or those taking medications for diabetes, the association between the TyG index and ESKD may be weak. A previous study on CKD progression in type 2 diabetes suggested that the association between the TyG index and CKD was more pronounced in the early stages of the disease (eGFR >60 mL/min/1.73 m²) and in individuals without a history of oral diabetes medications [32]. Other studies have reported that while the TyG index correlates significantly with microalbuminuria, its association with CKD [33] or eGFR [34] is less evident. These findings suggest that the TyG index has a stronger influence in the early stages of CKD but a diminished effect in later stages, such as ESKD. Insulin resistance, which is characterized by hyperglycemia and dysfunctional lipid oxidation and usage, contributes to early glomerular damage in patients with diabetes. However, as dysglycemia progresses, multifactorial processes, including inflammation, oxidative stress, and metabolic acidosis, exacerbate kidney damage [14,33]. Third, among young adults with MetS or those taking medications for dyslipidemia, MetS and dyslipidemia status can change over time due to lifestyle modifications, medications, or illness [35], potentially weakening the association between the TyG index and ESKD risk during follow-up. Additionally, as the 2 components of the TyG index (ie, fasting glucose and triglyceride levels) overlap with the 5 diagnostic criteria for MetS, this may diminish the effect of the TyG index in patients with MetS. Fourth, in high-risk populations with established metabolic abnormalities (eg, diabetes, MetS, or abnormal glucose metabolism), the relationship between the TyG index and risk of CKD may be nonlinear or modified by concomitant therapies. A restricted cubic spline analysis from a cohort of hypertensive patients with abnormal glucose metabolism demonstrated a U-shaped association between the TyG index and incident CKD (eGFR <60 mL/min/1.73 m² and/or proteinuria). The risk of CKD fell first and started to increase after the TyG index became higher than 8.94. Consistently, the risk of low eGFR also appeared to decrease first and then increase as the TyG quartile increased [36]. Taken together, these data partially support our finding that the TyG index predicted ESKD development mainly in metabolically healthier young adults, whereas its prognostic value was attenuated in those with preexisting diabetes or MetS, or those on medications for diabetes or dyslipidemia. Such patients have relatively higher baseline renal risk, and progression to ESKD is strongly influenced by disease duration, glycemic control, and renoprotective medications such as renin-angiotensin system inhibitors. Further studies are needed to elucidate the association between the TyG index and risk of ESKD in metabolically unhealthy young adults.

This study had several limitations. First, its observational design and limited scope of the database prevented the inclusion of various confounding factors that contribute to the pathogenesis of ESKD, such as inflammation, nutrition, comorbidities, other kidney diseases, and additional metabolic risk factors. Particularly, the lack of consideration of genetic or congenital kidney diseases is a limitation in this cohort of young adults. Second, as the TyG index does not adequately reflect changes in response to drug treatment or lifestyle modification, the evidence supporting its role in the development of ESKD remains limited. Third, the main analysis did not adjust for other metabolic risk factors due to the lack of such information in the NHIS database. Fourth, there is no biological validation supporting the association between the TyG index and ESKD. Finally, the insignificant association between the TyG index and ESKD in subgroup analyses of young adults with diabetes, MetS, and those taking medications for diabetes or dyslipidemia is also a limitation that needs further research. Despite these limitations, the findings of our study remain significant, as the study was based on a large population-based cohort with a longitudinal database. The study population uniquely focused on young adults, a group that has been underrepresented in ESKD research. Additionally, adjustments were made for established ESKD risk factors, eGFR, and income, and the TyG index was examined in quartiles and deciles, using follow-up and time-averaged values, enabling a comprehensive evaluation of its association with ESKD risk.

In conclusion, this study demonstrated that a higher TyG index is significantly associated with an increased risk of incident ESKD in young adults. These findings suggest that the TyG index may have potential utility in identifying young adults at risk of developing ESKD. Further comprehensive studies are warranted to confirm the relationship and clarify the role of the TyG index in the development of ESKD in young adult populations.