Research report: Associations of triglyceride-glucose-obesity composite indices with all-cause and cardiovascular mortality in individuals with cardiovascular-kidney-metabolic syndrome stages 0–3 and their potential incremental value beyond a PREVENT-variable-based model

A large analysis of United States adults with early cardiovascular-kidney-metabolic syndrome has found that seven triglyceride-glucose indices, alone and combined with measures of obesity, were independently associated with death from any cause, but none of them showed a significant association with death from cardiovascular disease once full statistical adjustments were applied. The study, published in Cardiovascular Diabetology, drew on 12,947 adults classified in cardiovascular-kidney-metabolic syndrome stages 0 to 3 within the National Health and Nutrition Examination Survey, with a median follow-up of 10.25 years. During that period, 1,284 deaths from any cause and 342 cardiovascular deaths occurred. The strongest association with all-cause mortality belonged to the TyG-BMI index, which showed a hazard ratio of 1.99 with a 95 percent confidence interval of 1.21 to 3.28, followed closely by TyG-BRI at a hazard ratio of 1.89 with a confidence interval of 1.38 to 2.59.

The investigation was framed around a recent shift in cardiovascular risk assessment. The PREVENT equations, developed to estimate cardiovascular risk, represent what the study authors describe as a paradigm shift because they incorporate cardiovascular-kidney-metabolic health metrics directly into risk prediction. This raised a specific and unresolved question: in people already classified within cardiovascular-kidney-metabolic syndrome stages 0 to 3, do the triglyceride-glucose index and its obesity composite variants carry independent associations with all-cause and cardiovascular mortality, and do they add measurable predictive value on top of a model built from PREVENT variables? Because these inputs come from simple blood tests and basic body measurements, the indices have attracted attention as low-cost markers that could, in principle, flag risk that standard models miss.

To answer the question, the researchers calculated seven indices for each participant: TyG alone and six TyG-obesity composite indices. Associations with mortality were assessed using weighted Cox proportional hazards models, which are the standard tool for estimating how a exposure relates to the rate of an event over time in survey data with complex sampling. Cumulative incidence of all-cause mortality was displayed with Kaplan-Meier curves, while cardiovascular mortality, being a competing event alongside other causes of death, was analyzed with Aalen-Johansen cumulative incidence functions and with both cause-specific Cox and Fine-Gray models, the latter explicitly accounting for the competing risk of non-cardiovascular death. Restricted cubic splines were used to test whether the relationships were linear or followed a nonlinear curve, and a Cox-based quantile g-computation approach estimated the joint effect of all seven indices considered together. The analysis battery also included subgroup, stage-specific, and sensitivity analyses, reflecting the authors’ effort to test whether the findings were robust across different slices of the population.

The central result concerned all-cause mortality. Kaplan-Meier curves showed that the cumulative incidence of death from any cause rose across quartiles of the TyG index, and the same graded pattern appeared for cardiovascular mortality in the Aalen-Johansen curves. After full adjustment for confounders, all seven indices remained independently associated with all-cause mortality, with TyG-BMI and TyG-BRI showing the strongest associations. The picture for cardiovascular mortality was strikingly different. In the fully adjusted Model 3, none of the seven indices retained a statistically significant association with cardiovascular death, and the competing-risk analyses using both cause-specific Cox and Fine-Gray models likewise showed no significant associations. In analyses stratified by quartile, one exception appeared: TyG-WWI was associated with increased cardiovascular mortality risk in the third quartile, with a hazard ratio of 4.53 and a confidence interval of 1.41 to 14.58, but this association was not present in the fourth quartile, a pattern that complicates any simple interpretation of a dose-response relationship.

Stage-specific analyses added further texture to the results. The associations between the indices and all-cause mortality were mainly significant among individuals in cardiovascular-kidney-metabolic syndrome stage 2, suggesting that the value of these markers may concentrate in a particular segment of the CKM spectrum rather than uniformly across stages 0 through 3. For cardiovascular mortality, significant interactions with CKM stage were observed for TyG-BMI, TyG-WC, and TyG-WHtR, indicating that the relationship between these indices and cardiovascular death differed depending on the stage of syndrome. The restricted cubic spline analysis suggested that the associations of TyG-BMI and TyG-BRI with all-cause mortality were nonlinear rather than following a straight-line dose-response pattern. The quantile g-computation analysis, which pooled the seven indices, indicated a modest joint effect on cardiovascular mortality. Sex emerged as a significant effect modifier for several indices, with a statistical significance level below 0.001, meaning that the strength of the associations differed between men and women for those measures.

The most consequential part of the study concerned incremental predictive value, and here the results were sobering. When the TyG indices were added to a model built from PREVENT variables, the increases in the time-dependent area under the curve at 10 years were all below 0.01, a threshold that represents a very small gain in discrimination. Bootstrap resampling analyses showed small but statistically significant improvements in the C-statistic for two indices in relation to all-cause mortality: TyG-ABSI produced a change of 0.0024 with a confidence interval of 0.0006 to 0.0055, and TyG-WWI produced a change of 0.0017 with a confidence interval of 0.0002 to 0.0046. No such improvement appeared for cardiovascular mortality. Survival-based net reclassification improvement and integrated discrimination improvement showed modest gains only for TyG-ABSI, with an NRI of 0.092 and an IDI of 0.003, and TyG-WWI, with an NRI of 0.094 and an IDI of 0.003, and again only for all-cause mortality. Calibration and decision curve analyses were part of the evaluation framework, and the overall conclusion the authors drew was that incremental predictive value beyond the PREVENT-variable-based model was limited, with modest improvement in risk reclassification for all-cause mortality only.

The robustness checks largely supported the main findings. Results from multiple imputation sensitivity analyses, which address missing data by statistically imputing plausible values, were consistent with the complete-case analysis.

What the results mean in practice is a question the authors addressed with caution. As low-cost indicators that rely only on routine laboratory tests and simple anthropometric measurements, the TyG and TyG-obesity composite indices may, in the authors’ words, help identify individuals at underestimated risk for mortality when used alongside a PREVENT-variable-based model. However, they emphasized that the net clinical benefit of such use still requires further validation in prospective studies. The distinction between statistical association and clinical usefulness runs through the entire paper: an index can be independently associated with mortality in a cohort and still add so little discrimination or reclassification beyond an existing model that its routine deployment would change few clinical decisions. The modest reclassification gains observed here, confined to all-cause mortality and to just two of the seven indices, illustrate that gap precisely.

The study’s scope also deserves emphasis. The population consisted of adults in cardiovascular-kidney-metabolic syndrome stages 0 to 3 drawn from a United States national survey, and the strongest associations with all-cause mortality were concentrated in stage 2, so the findings may not generalize uniformly to other populations or to stage 4 cardiovascular-kidney-metabolic disease. The authors concluded that in this population the TyG and TyG-obesity composite indices were independently associated with all-cause mortality, with TyG-BMI and TyG-BRI showing the strongest associations, that no significant associations were observed with cardiovascular mortality, and that incremental predictive value beyond a PREVENT-variable-based model was limited, with modest improvement in risk reclassification for all-cause mortality only. Their final assessment was that the clinical utility of these indices warrants further validation, a measured conclusion that matches the evidence rather than outrunning it.

Subject of Research: Associations of triglyceride-glucose and TyG-obesity composite indices with all-cause and cardiovascular mortality in adults with cardiovascular-kidney-metabolic syndrome stages 0-3

Article Title: Associations of triglyceride-glucose-obesity composite indices with all-cause and cardiovascular mortality in individuals with cardiovascular-kidney-metabolic syndrome stages 0–3 and their potential incremental value beyond a PREVENT-variable-based model

Article References: Wang, H., Yang, C., Ge, S., Jin, R., Wang, X., Hao, Z., Wang, X., Zhao, M., He, H., & Tian, F. (2026). Associations of triglyceride-glucose-obesity composite indices with all-cause and cardiovascular mortality in individuals with cardiovascular-kidney-metabolic syndrome stages 0–3 and their potential incremental value beyond a PREVENT-variable-based model. Cardiovascular Diabetology. https://doi.org/10.1186/s12933-026-03427-2

Image Credits: AI Generated

DOI: 10.1186/s12933-026-03427-2

Keywords: TyG index, TyG-BMI, TyG-BRI, cardiovascular-kidney-metabolic syndrome, all-cause mortality, cardiovascular mortality, PREVENT equations, NHANES, risk prediction, C-statistic, net reclassification improvement, obesity indices