The National Health and Nutrition Examination Survey (NHANES) is a research program led by the U.S. Centers for Disease Control and Prevention (CDC) to assess the health status of adults and children nationwide. Initiated in 1960, it has operated as a continuous program since 1999, surveying approximately 5,000 individuals annually. The survey collects comprehensive data on demographics, socioeconomic status, diet, and health (14, 15). This study utilized data from 10 NHANES cycles spanning 1999 to 2018. Additionally, NHANES data were linked with the National Death Index (NDI) to obtain follow-up information, including follow-up duration, survival status, and causes of death (16).

The TyG index is calculated by taking the natural logarithm of the product of fasting triglyceride and glucose levels for each participant. Systemic inflammation indicators are derived from complete blood counts. TyG-NLR is obtained by multiplying the TyG index by various systemic inflammation indicators (17). The formulas for these calculations are as follows:

The NHANES dataset provided all variables used in this study, including gender, age, race, education level, and socioeconomic status. Other variables included smoking status, alcohol consumption, medical history, medication use, body mass index (BMI), complete blood counts, and biochemical blood tests. Dyslipidemia was defined using multiple parameters: low-density lipoprotein cholesterol (LDL-C) ≥ 3.37 mmol/L, total cholesterol (TC) ≥5.18 mmol/L, triglycerides (TG) ≥ 1.7 mmol/L, and high-density lipoprotein cholesterol (HDL-C) <1.04 mmol/L for men or < 1.30 mmol/L for women (18). The use of lipid-lowering medications was also considered in diagnosing dyslipidemia. Hypertension was defined by an average systolic blood pressure ≥ 140 mmHg and/or diastolic blood pressure ≥ 90 mmHg (19). Diabetes was diagnosed based on physician confirmation, fasting blood glucose ≥ 7.0 mmol/L, HbA1c ≥ 6.5%, or current use of antidiabetic medications (20). Smoking status was classified into never smokers (fewer than 100 cigarettes in a lifetime), former smokers (previously smoked but quit), and current smokers (≥ 100 cigarettes in a lifetime and currently smoking) (21). Alcohol consumption was categorized as heavy drinking (≥ 3 drinks/day for women or ≥ 4 drinks/day for men, or ≥ 5 days/month), moderate drinking (≤ 2 drinks/day for women or ≤ 3 drinks/day for men, or ≥ 2 days/month), light drinking (below moderate or heavy drinking levels), and non-drinkers (lifetime consumption < 12 drinks) (22). Medication use and cancer history were determined via survey responses (23).

All analyses were performed using R software (version 4.2.2), with statistical significance set at P < 0.05. Categorical variables are presented as frequencies and percentages, while continuous variables are reported as medians with interquartile ranges (IQR). The Kaplan-Meier method was used to estimate survival probabilities with the log-rank test. We employed three models to assess the impact of potential confounders. Model 1 was unadjusted, Model 2 adjusted for sex and age, and Model 3 further adjusted for race, PIR, education, BMI, smoking status, alcohol consumption, hypertension, diabetes, dyslipidemia, cancer, lipid-lowering medications, ALT, and AST. A restricted cubic spline (RCS) with four knots was used to explore the dose-response relationship between TyG-NLR and mortality.

This study included a total of 15,388 participants, with 52.48% being female ( Figure 1 ). During a median follow-up period of 118months, 2,333 participants experienced outcome events. Table 1 outlines the baseline characteristics of participants in the survivor and non-survivor groups, highlighting significant demographic, clinical, and biochemical differences. Compared to survivors, non-survivors were significantly older, with 69.19% being 60 years or older, compared to 22.35% in the survivor group. They were more likely to be male (38.36% vs. 49.15%) and predominantly non-Hispanic White (63.20% vs. 61.08%). Socioeconomic disparities were evident, with non-survivors having a higher proportion of individuals with a PIR < 1 (13.84% vs. 7.65%) and lower education levels, with only 17.66% having achieved more than high school education compared to 60.99% among survivors. Clinically, non-survivors exhibited a higher burden of comorbidities, including hypertension (93.78% vs. 69.64%), hyperlipidemia (39.15% vs. 30.94%), diabetes, and cancer (37.96% vs. 16.43%). Additionally, a higher proportion of non-survivors were on lipid-lowering medications (19.48% vs. 18.37%). These findings suggest a higher prevalence of chronic conditions and associated cardiovascular risks among non-survivors. Biochemically, non-survivors exhibited higher levels of fasting blood glucose (5.87 vs. 5.40 mmol/L), triglycerides (1.35 vs. 1.08 mmol/L), LDL-C (2.90 vs. 2.58 mmol/L), and aspartate aminotransferase (21.00 vs. 19.00 U/L). They also displayed elevated markers of inflammation, with higher levels of white blood cells (6.70 vs. 6.20 10^9/L), neutrophils (4.00 vs. 3.30 10^9/L), and monocytes (0.60 vs. 0.40 10^9/L), alongside lower lymphocyte levels (1.70 vs. 1.90 10^9/L), indicating a heightened inflammatory state and potential immune suppression.

Kaplan-Meier survival analysis adjusted for all covariates based on Model 3 showed significant differences in long-term survival outcomes across groups with different TyG index levels ( Figure 2A , P< 0.001). A similar pattern was observed for the TyG-NLR index ( Figure 2B , P< 0.001). We employed three models to evaluate the association between the TyG and TyG-NLR indices and all-cause mortality ( Table 2 , Table 3 ). In Model 3, no significant association was observed between the TyG index and all-cause mortality across the quartiles, with hazard ratios (HRs) ranging from 0.79 (95% confidence interval [CI]: 0.51–1.15) in the third quartile to 1.02 (95% CI: 0.62–1.48) in the highest quartile ( Table 4 ). In contrast, the TyG-NLR index demonstrated a significant positive association with all-cause mortality across all models. In Model 3, participants in the highest quartile of the TyG-NLR index had a significantly higher risk of all-cause mortality compared to those in the lowest quartile (HR = 1.63, 95% CI: 1.08–2.54). These findings suggest that while the TyG index alone may not robustly predict mortality risk, the combined TyG-NLR index effectively captures the synergistic impact of insulin resistance and systemic inflammation, making it a more reliable predictor of all-cause mortality.

After adjusting for potential confounders, we analyzed the dose-response relationship between the TyG and TyG-NLR indices and mortality using RCS analysis (consistent with Model 3). The dose-response relationships between both indices and all-cause mortality were nonlinear (P for nonlinearity < 0.05), as shown in Figure 3 . When assessing the dose-response relationships between the TyG and TyG-NLR indices and mortality ( Figure 3 ), a similar pattern emerged. Notably, regardless of the specific shape of the dose-response curve, once the respective threshold points were exceeded, an increase in both the TyG and TyG-NLR indices was associated with a higher risk of mortality.

The strengths of this study lie in it being the first to propose the TyG-NLR index and confirm its association with mortality in the general population. Additionally, unlike previous studies that primarily focused on populations with metabolic disorders, our study demonstrates that the TyG-NLR index can serve as a predictor of mortality risk in the general population. However, this study has several limitations. First, as an observational cohort study, we cannot establish a direct causal relationship between the TyG-NLR index and mortality. Furthermore, the underlying mechanisms driving the significant association between the TyG-NLR index and mortality remain unclear. Lastly, since the study participants were from the United States, further validation is needed across different populations and regions.