Abstract
Background
The triglyceride glucose (TyG) index, a surrogate marker of insulin resistance, has been linked to cardiovascular risk. However, its prognostic role in critically ill patients with atherosclerotic cardiovascular disease (ASCVD) remains unclear.
Methods
In this retrospective cohort study using the MIMIC-IV database, we identified 2,493 ASCVD patients and stratified them into TyG tertiles upon Intensive Care Unit (ICU) admission. Primary outcomes were 30-, 90-, and 365-day mortality. Secondary outcomes included the use and timing of mechanical ventilation (MV) and vasopressors. Cox regression, restricted cubic spline, and Fine–Gray competing risk models were applied. Propensity score matching (PSM) was performed as a sensitivity analysis. Subgroup analyses were conducted by age, sex, race, hypertension, diabetes, and statin use. Mediation analysis using white blood cell (WBC) count explored systemic inflammation.
Results
Higher TyG tertiles were consistently associated with increased mortality: 30-day (HR 1.85, 95% CI 1.46–2.36, P < 0.001), 90-day (HR 1.73, 95% CI 1.39–2.15, P < 0.001), and 365-day (HR 1.66, 95% CI 1.37–2.03, P < 0.001). These associations remained robust in propensity score-matched (PSM) analyses. Elevated TyG was also linked to greater need for MV (OR 1.93, 95% CI 1.66–2.24, P < 0.001) and vasopressor support (OR 1.58, 95% CI 1.35–1.86, P < 0.001) within the first 24 h, and to earlier initiation of these interventions in competing risk models (MV): sHR (subdistribution hazard ratio) 1.46, 95% CI 1.35–1.57, P < 0.001; vasopressor use: sHR 1.41, 95% CI 1.30–1.53, P < 0.001. WBC count partially mediated the TyG-mortality association, accounting for 22.18% of the total effect (P < 0.001) for 30-day mortality. In exploratory subgroup analyses, the association between TyG and mortality appeared more pronounced in nondiabetic patients (30-day HR 1.58, 95% CI 1.35–1.84) than in diabetic patients (HR 1.13, 95% CI 0.94–1.36), although these findings should be interpreted cautiously.
Conclusion
In critically ill patients with ASCVD, higher TyG index was associated with adverse outcomes, including increased mortality and earlier initiation of organ-supportive interventions. These findings suggest that TyG could serve as a practical biomarker for early risk stratification in this high-risk population, though prospective studies are warranted to confirm its clinical utility.
Graphical Abstract
Supplementary Information
The online version contains supplementary material available at 10.1186/s40001-026-03994-w.
Keywords: Triglyceride glucose index, Atherosclerotic cardiovascular disease, Mortality, Inflammation, Insulin resistance, Critical care
Introduction
Cardiovascular disease (CVD) remains the leading cause of death and disability worldwide, with atherosclerotic cardiovascular disease (ASCVD)—including ischemic heart disease and ischemic stroke—accounting for the majority of CVD-related mortality [1–5]. Despite advances in lipid-lowering therapies and interventional strategies, early prognostic assessment in critically ill ASCVD patients remains challenging [6, 7]. Identifying simple and reliable biomarkers is therefore crucial for improving outcomes in this high-risk population.
Insulin resistance (IR) plays a pivotal role in cardiovascular pathogenesis, and the triglyceride glucose (TyG) index is a validated surrogate of IR [8–10]. Compared with other IR indices such as HOMA-IR, the TyG index requires only fasting triglyceride and glucose levels, which are routinely measured in Intensive Care Unit (ICU) settings, making it a simple and feasible biomarker for critically ill patients. Several studies using the MIMIC database have examined the TyG index in specific ASCVD subtypes. Zhang et al. [11] and Chen et al. [12] reported that elevated TyG was associated with higher mortality in patients with coronary artery disease, while Cai et al. [13] demonstrated a similar association in critically ill patients with ischemic stroke. More recently, Wang et al. [14] further highlighted the prognostic relevance of TyG in coronary artery disease cohorts. However, these investigations primarily focused on mortality within a single ASCVD subtype, whereas other clinically meaningful ICU endpoints such as early initiation of MV or vasopressor therapy have rarely been evaluated.
In this retrospective cohort study using the MIMIC-IV database, we investigated the association between the TyG index and clinical outcomes, including mortality and early use of MV and vasopressors, in critically ill patients with ASCVD. We also evaluated whether systemic inflammation, reflected by white blood cell (WBC) count, mediates this relationship. These findings aim to inform clinical decision-making and enhance early risk identification in intensive care settings.
Materials and methods
Study design and population
This study was conducted using data from the MIMIC-IV v3.1, an open-access critical care database maintained by the Massachusetts Institute of Technology Laboratory for Computational Physiology [15]. The database contains de-identified health records of over 60,000 adult first-time ICU admissions at the Beth Israel Deaconess Medical Center between 2008 and 2022 [16]. Data access was granted following completion of the required training modules (Certification ID: 65444380) and compliance with institutional data use agreements. The reporting of this study and its design conformed to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines [17].
From the MIMIC-Ⅳ database, 94,458 patient records were extracted, with only the first ICU admission considered for those with multiple admissions. Among 65,366 initial ICU admissions, 15,468 met criteria for ASCVD, identified using previously validated ICD-9-CM and ICD-10 codes for ischemic stroke (ICD-9-CM 433.xx/434.xx; ICD-10 I63.x, I65–I66) and ischemic heart disease (ICD-10 I20–I25; ICD-9-CM 410–414); full code lists are provided in Supplementary Table 1. Exclusion criteria were age < 18 years (0 cases), ICU stay < 24 h (n = 2,175), and missing triglyceride or fasting glucose data on the first day (n = 10,800). Finally, 2,493 patients were enrolled (Fig. 1). ASCVD patients were stratified into TyG tertiles: T1 (n = 831, reference), T2 (n = 831), and T3 (n = 831). To evaluate potential selection bias, baseline characteristics were compared between included and excluded patients (Supplementary Table 4).
Fig. 1.
Flowchart for participants from the MIMIC-IV (v 3.1). MIMIC-IV Medical Information Mart for Intensive Care IV, ASCVD atherosclerotic cardiovascular disease, ICU intensive care unit
Variable extraction
Data were extracted from the MIMIC-IV database using PostgreSQL v17.6.1 and Navicat Premium v17.2.5 via structured SQL queries. The dataset comprised demographic characteristics (age, sex, race, weight and height), vital signs, comorbidities, clinical interventions, scoring systems, laboratory parameters, and clinical outcomes. Within 24 h of ICU admission, all baseline measurements were obtained. To minimize bias from missing data, variables with > 40% missing values (e.g., height, 55.39%) were excluded given the high uncertainty of imputation. For variables with 1–15% missing values, multiple imputation (MI) with 5 imputed datasets was performed, and pooled estimates were reported [18]. The proportions of missing data for each variable are presented in Supplementary Table 2. Potential confounders were selected a priori based on systematic literature review and clinical relevance. Multicollinearity among independent variables was assessed using variance inflation factors (VIFs), with VIF values < 5 indicating no significant multicollinearity (Supplementary Table 3).Table: Please specify the significance of the [Bold emphasis] reflected inside Table [1] by providing a description in the form of a table footnote. Otherwise, kindly amend if deemed necessary.Added a table footnote to clarify the bold emphasis: “Bold values indicate statistical significance (P < 0.05).”
TyG index calculation
To minimize confounding by therapeutic interventions, baseline triglyceride and fasting blood glucose levels measured within the first 24 h of ICU admission were used for TyG index calculation. The TyG index, the primary exposure variable, was calculated as ln [triglycerides (mg/dL) × fasting blood glucose (mg/dL)/2]. The TyG index is a dimensionless, logarithmic composite indicator that reflects insulin resistance. In clinical and research settings, the TyG index is typically analyzed as a continuous variable, with values in adult populations commonly ranging from approximately 7.0–11.0, depending on the study population; higher values are associated with greater insulin resistance and metabolic dysregulation.
Outcome
The primary outcomes focused on mortality at 30, 90, and 365 days following ICU admission. Secondary analyses assessed the use of vasopressor therapy and MV within the first 24 h of ICU admission, as well as the time from admission to the first initiation of these interventions within 28 days.
Statistical analysis
Continuous variables are summarized as mean ± standard deviation (SD) if normally distributed, or as median with interquartile range (IQR) if skewed. Categorical variables are expressed as counts and percentages. Group comparisons were performed using the chi-square test or Fisher’s exact test for categorical variables, one-way ANOVA for normally distributed continuous variables, and the Kruskal–Wallis H test for nonparametric data.
Survival differences among TyG tertiles were assessed using Kaplan–Meier curves with the log-rank test. To investigate the association between TyG levels and 30-, 90-, and 365-day mortality, Cox proportional hazards models were applied with adjustment for potential confounders. Restricted cubic spline (RCS) models were used to explore potential nonlinear relationships between TyG and mortality. To address residual confounding, propensity score matching (PSM) was performed using multinomial logistic regression including demographic, clinical, and laboratory covariates. Patients were matched 1:1 using nearest-neighbor matching with a caliper of 0.2. This yielded 753 matched patients (251 per tertile). Covariate balance was evaluated using standardized mean differences (SMDs), with SMD < 0.1 indicating good balance. Multivariable Cox models were then re-applied in the matched cohort. For secondary outcomes, logistic regression was used to assess the likelihood of mechanical ventilation (MV) and vasopressor therapy within the first 24 h of ICU admission, and linear regression was used to analyze ventilator-free and vasopressor-free days within 28 days. Fine–Gray competing risk models were additionally applied to account for death or ICU discharge. Results are reported as odds ratios (ORs), regression coefficients (β), or subdistribution hazard ratios (sHRs) with 95% CIs, as appropriate. Subgroup analyses were performed to evaluate potential effect modification across major demographic and clinical strata, including age, sex, race, hypertension, diabetes, and statin use. A post hoc exploratory mediation analysis was performed using WBC count (measured within 24 h of ICU admission) to examine whether systemic inflammation might partially mediate the association between TyG and mortality; given the simultaneous measurement, temporal sequence cannot be inferred. To further assess the potential influence of unmeasured confounding, we performed E-value analysis for the association between TyG and mortality.
Statistical analyses were performed using R4.2.2 (http:// www.R-project.org, The R Foundation) and Free Statistics software version 2.1.1.
Results
Baseline demographic and clinical characteristics
A total of 2,493 patients diagnosed with ASCVD were enrolled in this study, comprising 1,112 females (44.6%) and 1,381 males (55.4%), with a mean age of 69.8 years. Participants were categorized into three groups according to TyG tertiles: T1 (TyG < 8.57), T2 (8.57 ≤ TyG < 9.15), and T3 (TyG ≥ 9.15). Baseline characteristics for each group are summarized in Table 1. Compared with the lower TyG categories, individuals in the highest tertile exhibited elevated levels of heart rate, respiratory rate, anion gap, blood urea nitrogen, presence of sepsis, acute kidney injury, weight, body mass index, serum potassium, vasopressin administration, white blood cell count, and creatinine. All evaluated scoring systems (SOFA, SAPS II, and OASIS) revealed progressively deteriorating conditions across higher tertile groups. In contrast, most baseline variables did not differ significantly across groups (P > 0.05).
Table 1.
Baseline characteristics and outcomes of participants classified by TyG index tertiles
| Variables | Total | TyG index | P-value | ||
|---|---|---|---|---|---|
| (n = 2493) | T1 (n = 831) | T2 (n = 831) | T3 (n = 831) | ||
| Demographic | |||||
| Age, years | 69.8 ± 14.9 | 72.6 ± 15.4 | 70.6 ± 14.2 | 66.1 ± 14.2 | < 0.001 |
| Sex, male, n (%) | 1381 (55.4) | 435 (52.3) | 477 (57.4) | 469 (56.4) | 0.089 |
| Race, White, n (%) | 1375 (55.2) | 469 (56.4) | 473 (56.9) | 433 (52.1) | 0.094 |
| Weight, (kg) | 81.9 ± 23.2 | 76.5 ± 20.8 | 80.2 ± 21.6 | 89.0 ± 25.1 | < 0.001 |
| Height, (cm) | 169.3 ± 10.4 | 168.5 ± 10.6 | 169.0 ± 10.5 | 170.0 ± 10.2 | 0.134 |
| BMI, kg/m2 | 29.3 ± 7.7 | 27.0 ± 5.8 | 28.6 ± 7.8 | 31.4 ± 8.1 | < 0.001 |
| Vital signs | |||||
| Heart rate (beats/min) | 80.4 ± 15.5 | 77.6 ± 14.5 | 80.2 ± 15.5 | 83.4 ± 15.8 | < 0.001 |
| MAP, mmHg | 87.7 ± 12.6 | 88.4 ± 12.4 | 88.0 ± 12.5 | 86.8 ± 12.9 | 0.023 |
| Resp rate (beats/min) | 19.5 ± 3.6 | 18.7 ± 2.9 | 19.4 ± 3.4 | 20.3 ± 4.2 | < 0.001 |
| Spo2 (%) | 96.7 ± 2.0 | 96.7 ± 1.9 | 96.7 ± 2.0 | 96.8 ± 2.0 | 0.419 |
| Comorbidities | |||||
| CHF, n (%) | 762 (30.6) | 224 (27) | 251 (30.2) | 287 (34.5) | 0.003 |
| Renal disease, n (%) | 452 (18.1) | 120 (14.4) | 138 (16.6) | 194 (23.3) | < 0.001 |
| Malignant cancer, n (%) | 165 (6.8) | 44 (5.3) | 65 (7.8) | 56 (6.7) | 0.115 |
| Hypertension, n (%) | 1911 (76.7) | 614 (73.9) | 634 (76.3) | 663 (79.8) | 0.017 |
| AF, n (%) | 926 (37.1) | 342 (41.2) | 327 (39.4) | 257 (30.9) | < 0.001 |
| Diabetes, n (%) | 886 (35.5) | 157 (18.9) | 248 (29.8) | 481 (57.9) | < 0.001 |
| Sepsis, n (%) | 958 (38.4) | 236 (28.4) | 295 (35.5) | 427 (51.4) | < 0.001 |
| AKI, n (%) | 1826 (73.2) | 532 (64.0) | 617 (74.2) | 677 (81.5) | < 0.001 |
| COPD, n (%) | 218 (8.7) | 60 (7.2) | 83 (10.0) | 75 (9.0) | 0.128 |
| Clinical treatment | |||||
| Betablockers, n (%) | 1035 (42.1) | 353 (43.1) | 375 (45.8) | 307 (37.5) | 0.002 |
| Statin, n (%) | 1135 (46.2) | 384 (46.9) | 379 (46.3) | 372 (45.4) | 0.835 |
| Vent1day, n (%) | 605 (24.6) | 120 (14.7) | 182 (22.2) | 303 (37.0) | < 0.001 |
| Vaso1day, n (%) | 417 (17.0) | 74 (9.0) | 138 (16.9) | 205 (25.0) | < 0.001 |
| Scoring system | |||||
| SOFA | 3.0 (1.0, 5.0) | 2.0 (1.0, 4.0) | 3.0 (1.0, 5.0) | 4.0 (2.0, 7.0) | < 0.001 |
| OASIS | 32.0 (26.0, 39.0) | 30.0 (25.0, 36.0) | 32.0 (26.0, 39.0) | 35.0 (28.0, 42.0) | < 0.001 |
| SAPS II | 33.0 (25.0, 42.0) | 31.0 (25.0, 39.0) | 33.0 (25.0, 41.0) | 35.0 (26.0, 46.5) | < 0.001 |
| Laboratory parameters | |||||
| Sodium, mmol/L | 137.7 ± 4.7 | 137.9 ± 4.7 | 137.9 ± 4.2 | 137.2 ± 5.2 | 0.002 |
| Potassium, mmol/L | 4.2 ± 0.7 | 4.1 ± 0.6 | 4.1 ± 0.7 | 4.2 ± 0.7 | < 0.001 |
| Chloride, mmol/L | 102.0 ± 5.4 | 102.6 ± 5.2 | 102.1 ± 4.9 | 101.2 ± 5.9 | < 0.001 |
| Calcium, mmol/L | 8.5 ± 0.8 | 8.5 ± 0.7 | 8.5 ± 0.8 | 8.3 ± 0.9 | < 0.001 |
| Bicarbonate, mmol/L | 21.4 ± 4.4 | 22.0 ± 3.6 | 21.7 ± 4.2 | 20.3 ± 5.0 | < 0.001 |
| Hemoglobin, g/L | 11.6 ± 2.3 | 11.7 ± 2.2 | 11.7 ± 2.3 | 11.3 ± 2.4 | < 0.001 |
| WBC, 109 /L | 11.3 ± 5.7 | 9.7 ± 4.3 | 11.1 ± 4.9 | 13.3 ± 7.0 | < 0.001 |
| Platelets, 109 /L | 197.0 (153.8, 246.2) | 193.0 (155.0, 242.0) | 203.0 (158.2, 249.0) | 194.0 (146.0, 247.5) | 0.038 |
| Creatinine, mg/dL | 1.0 (0.8, 1.4) | 0.9 (0.8, 1.2) | 1.0 (0.8, 1.3) | 1.1 (0.9, 1.9) | < 0.001 |
| INR | 1.2 (1.1, 1.3) | 1.2 (1.1, 1.3) | 1.2 (1.1, 1.3) | 1.2 (1.1, 1.4) | 0.267 |
| Anion gap, mmol/L | 15.9 ± 4.7 | 14.7 ± 4.0 | 15.6 ± 4.0 | 17.3 ± 5.6 | < 0.001 |
| Lactate, mmol/L | 2.0 (1.3, 3.4) | 1.8 (1.2, 2.8) | 1.9 (1.3, 3.1) | 2.3 (1.5, 4.1) | < 0.001 |
| Total bilirubin, mg/dL | 0.6 (0.4, 0.8) | 0.6 (0.4, 0.9) | 0.6 (0.4, 0.8) | 0.6 (0.4, 0.8) | 0.074 |
| Total cholesterol, mg/dL | 160.7 ± 49.9 | 151.2 ± 43.1 | 160.9 ± 48.6 | 172.2 ± 56.6 | < 0.001 |
| HDL, mg/dL | 47.6 ± 16.7 | 53.0 ± 17.3 | 47.5 ± 16.6 | 41.0 ± 13.1 | < 0.001 |
| Triglycerides, mg/dL | 143.7 ± 186.0 | 71.9 ± 21.0 | 111.9 ± 31.5 | 247.3 ± 292.4 | < 0.001 |
| Glucose, mg/dL | 127.0 (103.0, 167.0) | 105.0 (91.5, 124.0) | 125.0 (106.0, 152.0) | 176.0 (133.0, 245.5) | < 0.001 |
| Outcome | |||||
| 30-day mortality, n (%) | 489 (19.6) | 134 (16.1) | 151 (18.2) | 204 (24.5) | < 0.001 |
| 90-day mortality, n (%) | 607 (24.3) | 173 (20.8) | 193 (23.2) | 241 (29) | < 0.001 |
| 365-day mortality, n (%) | 750 (30.1) | 211 (25.4) | 254 (30.6) | 285 (34.3) | < 0.001 |
| Ventilator-free days at day 28 | 28.0 (23.9, 28.0) | 28.0 (27.3, 28.0) | 28.0 (25.6, 28.0) | 27.2 (0.0, 28.0) | < 0.001 |
| Vasopressor-free days at day 28 | 28.0 (25.9, 28.0) | 28.0 (28.0, 28.0) | 28.0 (26.8, 28.0) | 28.0 (8.1, 28.0) | < 0.001 |
Data are presented as the mean (SD), (IQR), or number (%), as appropriate. Bold values indicate statistical significance (P < 0.05).
BMI body mass index, MAP mean arterial pressure, CHF congestive Heart failure, AF atrial fibrillation, AKI acute kidney injury, COPD chronic pulmonary disease, SOFA sequential organ failure assessment, OASIS oxford acute severity of illness score, SAPS II simplified acute physiology score II, WBC white blood cells, INR international normalized ratio, HDL high-density lipoprotein
Comparison between included and excluded patients
To evaluate potential selection bias due to missing triglyceride or fasting glucose data, we compared baseline characteristics between patients included in the final cohort (n = 2,493) and those excluded (n = 10,801). Significant differences were observed in demographics, comorbidities, and laboratory variables, indicating that the analytic cohort may not fully represent the broader population of critically ill ASCVD patients (Supplementary Table 4).
Clinical outcomes
Univariate and multivariable Cox regression analyses were conducted to assess the association of the TyG index with 30-, 90-, and 365-day mortality in critically ill patients with ASCVD (Table 2). In fully adjusted models, when TyG was analyzed as a continuous variable, elevated TyG levels showed a significant association with increased mortality across all assessed time intervals: 30-day (hazard ratio [HR] 1.37, 95% confidence interval [CI] 1.22–1.54, P < 0.001), 90-day (HR: 1.35, 95% CI 1.22–1.51, P < 0.001), and 365-day (HR: 1.30, 95% CI 1.17–1.43, P < 0.001). In tertile analyses, patients in the highest TyG tertile (T3) had higher 30-day mortality vs. the lowest tertile (T1) (HR = 1.85; 95% CI 1.46–2.36; P < 0.001; P for trend < 0.001) Comparable associations were noted for 90- and 365-day mortality outcomes. RCS modeling (Fig. 2) revealed a linear dose–response relationship between TyG levels and mortality. These associations remained consistent in propensity score-matched analyses. After propensity score matching (n = 753; 251 per tertile), baseline characteristics were well balanced (all SMDs < 0.10; Supplementary Table 5). Cox models in the matched cohort showed results consistent with the primary analyses: compared with T1, T3 was associated with higher risks of 30-day (HR = 1.89; 95% CI 1.26–2.84; P = 0.001), 90-day (HR = 1.72; 95% CI 1.20–2.48; P < 0.001), and 365-day mortality (HR = 1.63; 95% CI 1.14–2.32; P = 0.007), while T2 vs. T1 was not significant. Results were consistent with Supplementary Fig. 2. For secondary outcomes, in fully adjusted logistic models, T3 had higher odds of early MV (OR = 2.87; 95% CI 2.17–3.81; P < 0.001) and vasopressor use (OR = 3.17; 95% CI 2.26–4.45; P < 0.001) within 24 h (Supplementary Table 6). Higher TyG was associated with fewer ventilator-free (β = −2.21) and vasopressor-free days (β = −1.98) within 28 days. Fine–Gray models indicated that each 1-unit increase in TyG was associated with a higher subdistribution hazard of earlier MV initiation (sHR = 1.46; 95% CI 1.35–1.57; P < 0.001) and vasopressors (sHR = 1.41; 95% CI 1.30–1.53; P < 0.001), after accounting for death and ICU discharge as competing events (Supplementary Table 7). Findings were consistent across modeling approaches.
Table 2.
Univariate and multivariate cox regression analyses of the association between TyG index and 30-, 90-, and 365-day mortality in patients with ASCVD
| Variable | Crude | Model I | Model II | Model III | ||||
|---|---|---|---|---|---|---|---|---|
| HR (95% CI) | P-value | HR (95% CI) | P-value | HR (95% CI) | P-value | HR (95% CI) | P-value | |
| 30-day | ||||||||
| TYG | 1.33 (1.20 ~ 1.47) | < 0.001 | 1.54 (1.39 ~ 1.71) | < 0.001 | 1.59 (1.42 ~ 1.78) | < 0.001 | 1.37 (1.22 ~ 1.54) | < 0.001 |
| Tertile | ||||||||
| T1 | 1 (Ref) | 1 (Ref) | 1 (Ref) | 1 (Ref) | ||||
| T2 | 1.15 (0.91 ~ 1.45) | 0.235 | 1.27 (1.01 ~ 1.61) | 0.042 | 1.28 (1.02 ~ 1.62) | 0.036 | 1.18 (0.93 ~ 1.49) | 0.180 |
| T3 | 1.62 (1.30 ~ 2.02) | < 0.001 | 2.09 (1.67 ~ 2.61) | < 0.001 | 2.24 (1.76 ~ 2.84) | < 0.001 | 1.85 (1.46 ~ 2.36) | < 0.001 |
| P for trend | < 0.001 | < 0.001 | < 0.001 | < 0.001 | ||||
| 90-day | ||||||||
| TYG | 1.29 (1.17 ~ 1.41) | < 0.001 | 1.50 (1.37 ~ 1.65) | < 0.001 | 1.53 (1.38 ~ 1.70) | < 0.001 | 1.35 (1.22 ~ 1.51) | < 0.001 |
| Tertile | ||||||||
| T1 | 1 (Ref) | 1 (Ref) | 1 (Ref) | 1 (Ref) | ||||
| T2 | 1.14 (0.93 ~ 1.40) | 0.201 | 1.27 (1.03 ~ 1.56) | 0.023 | 1.27 (1.04 ~ 1.57) | 0.022 | 1.18 (0.96 ~ 1.45) | 0.120 |
| T3 | 1.50 (1.23 ~ 1.82) | < 0.001 | 1.94 (1.58 ~ 2.36) | < 0.001 | 2.02 (1.63 ~ 2.50) | < 0.001 | 1.73 (1.39 ~ 2.15) | < 0.001 |
| P for trend | < 0.001 | < 0.001 | < 0.001 | < 0.001 | ||||
| 365-day | ||||||||
| TYG | 1.23 (1.13 ~ 1.34) | < 0.001 | 1.45 (1.33 ~ 1.58) | < 0.001 | 1.44 (1.31 ~ 1.58) | < 0.001 | 1.30 (1.17 ~ 1.43) | < 0.001 |
| Tertile | ||||||||
| T1 | 1 (Ref) | 1 (Ref) | 1 (Ref) | 1 (Ref) | ||||
| T2 | 1.24 (1.03 ~ 1.49) | 0.021 | 1.38 (1.15 ~ 1.66) | 0.001 | 1.37 (1.14 ~ 1.65) | 0.001 | 1.28 (1.06 ~ 1.54) | 0.009 |
| T3 | 1.46 (1.23 ~ 1.75) | < 0.001 | 1.90 (1.58 ~ 2.28) | < 0.001 | 1.89 (1.55 ~ 2.29) | < 0.001 | 1.66 (1.37 ~ 2.03) | < 0.001 |
| P for trend | < 0.001 | < 0.001 | < 0.001 | < 0.001 | ||||
Crude: Unadjusted
Model I: age + gender + race
Model II: age + gender + race + hypertension + diabetes + congestive heart failure + atrial fibrillation + renal disease + chronic pulmonary disease
Model III: age + gender + race + weight + hypertension + diabetes + congestive heart failure + atrial fibrillation + renal disease + chronic pulmonary disease + weight + heart rate + potassium + creatinine + hemoglobin + platelets + mean arterial pressure + betablockers + statins + international normalized ratio
Fig. 2.
Restricted cubic spline (RCS) curves illustrating the multivariable-adjusted hazard ratios for all-cause mortality at A 30 days, B 90 days, and C 365 days according to TyG index levels. The solid red lines represent hazard ratios derived from Cox regression models, adjusted for potential confounders including age, sex, race, weight, hypertension, diabetes, congestive heart failure, atrial fibrillation, renal disease, chronic pulmonary disease, heart rate, potassium, creatinine, hemoglobin, platelets, mean arterial pressure, betablockers, statin, and international normalized ratio. Shaded areas denote 95% confidence intervals. Knots were placed at the 5th, 35th, 65th, and 95th percentiles of TyG distribution. The horizontal dashed line represents a hazard ratio of 1.0 (reference), anchored at the median TyG value. The top and bottom 0.5% of TyG index were excluded to reduce the influence of extreme values
Kaplan–Meier survival curve
Kaplan–Meier analysis revealed significant differences at 30-, 90-, and 365-day mortality among the three groups, with fatality rates escalating incrementally across ascending the TyG tertiles (log-rank P < 0.001). The Kaplan–Meier curves are presented in Fig. 3.
Fig. 3.
Kaplan–Meier survival curves for all-cause mortality stratified by TyG tertiles. A 30-day mortality; B 90-day mortality; C 365-day mortality
Subgroup analysis
Stratified and interaction analyses were performed to assess the consistency of the association between the TyG index and 30-, 90-, and 365-day mortality across various subgroups. The stratified analyses considered age, sex, race, hypertension, diabetes, and statin use. Notably, significant effect modification by diabetes status was observed for all-cause mortality, with P-interaction values of 0.003 (30-day), 0.008 (90-day), and 0.003 (365-day), respectively. This subgroup analysis was post hoc and exploratory, not pre-specified. Among nondiabetic individuals, the TyG index showed a possible prognostic significance. Specifically, the HR for 30-day mortality was 1.58 (95% CI 1.35–1.84), compared with 1.13 (95% CI 0.94–1.36) in diabetic patients. Similar trends were observed across 90-day and 365-day mortality endpoints (Fig. 4).
Fig. 4.
Subgroup analyses of the association between TyG index and all-cause mortality at A 30 days, B 90 days, and C 365 days. Exploratory post hoc subgroup analyses were performed using multivariable-adjusted Cox regression models to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) across key clinical subgroups, including age, sex, race, hypertension, diabetes, and statin use. Adjustments were made for age, sex, race, weight, hypertension, diabetes, congestive heart failure, atrial fibrillation, renal disease, chronic pulmonary disease, heart rate, potassium, creatinine, hemoglobin, platelets, mean arterial pressure, betablockers, statin, and international normalized ratio. P-values for interaction were calculated to assess potential effect modification
Mediation analysis of the association between TyG index and ASCVD
Figure 5 summarizes the mediation analysis evaluating the role of white blood cell (WBC) count in the association between TyG index and all-cause mortality. In fully adjusted models, WBC analysis suggested significant mediation proportions of 22.18% for 30-day mortality (P < 0.001), with similar mediation effects observed for 90-day and 365-day outcomes. Elevated WBC count partially mediated the association between TyG and mortality, suggesting a role of systemic inflammation.
Fig. 5.
Mediation analysis of the association between TyG index and all-cause mortality at A 30 days, B 90 days, and C 365 days. The pathway diagrams illustrate the mediating effect of white blood cell (WBC) count on the association between TyG index and short-, intermediate-, and long-term mortality risk. The analysis was adjusted for age, sex, race, weight, hypertension, diabetes, congestive heart failure, atrial fibrillation, renal disease, chronic pulmonary disease, heart rate, potassium, creatinine, hemoglobin, platelets, mean arterial pressure, betablockers, statin, and international normalized ratio. All proportions and P-values for mediation are indicated in the diagrams, demonstrating that WBC partially mediates the relationship between TyG and mortality across different time frames
Sensitivity analysis
To assess the stability of the main results, we conducted several sensitivity analyses. First, after excluding individuals with a prior diagnosis of malignancy (Supplementary Table 8), the associations between the TyG index and 30-, 90-, and 365-day mortality were similar to the main analysis. Similar results were observed after excluding patients with liver failure or advanced renal insufficiency (Supplementary Table 9). For 30-day mortality, the E-value was 1.79, indicating that an unmeasured confounder would need to be strongly associated with both the TyG index and mortality to fully explain the observed association. These results are illustrated in Supplementary Fig. 1. In addition, propensity score-matched analyses produced estimates consistent with the multivariable models, with the highest TyG tertile remaining significantly associated with increased mortality. Baseline comparisons between included and excluded patients (Supplementary Table 4) indicated that excluded patients were generally older and had more comorbidities. For secondary outcomes, logistic regression models for early use of MV and vasopressors (Supplementary Table 7) were consistent with the main analysis, and Fine–Gray competing risk models (Supplementary Table 7) showed that higher TyG was associated with earlier initiation of these interventions after accounting for death and ICU discharge as competing events.
Discussion
This study, utilizing data from the MIMIC database, systematically evaluated the association between the TyG index and mortality outcomes in critically ill patients with ASCVD. Our analysis revealed that higher TyG indices were independently associated with an increased risk of both short- and long-term mortality. In addition to supporting its prognostic relevance, this study also revealed underlying mechanisms: elevated WBC counts partially accounted for the association between TyG and mortality, suggesting that systemic inflammation could be one explanatory pathway. Moreover, patients with higher TyG levels required MV and vasopressor support earlier and more frequently, indicating a more rapid decline in clinical status. These findings highlight the interplay between metabolism, inflammation, and timing of life-sustaining interventions. They also support the TyG index as a biomarker associated with mortality and a potential tool for early risk stratification.
Previous studies have highlighted the TyG index as a prognostic indicator across various clinical settings. In critically ill populations, elevated TyG levels have been correlated with increased mortality in sepsis, acute kidney injury, and hemorrhagic stroke [19–21], suggesting its utility as a simple and effective biomarker in intensive care units. Additionally, accumulating evidence has linked the TyG index to adverse cardiovascular outcomes, including heart failure, myocardial infarction, and cardiovascular mortality [22–25]. However, despite these advances, the prognostic value of the TyG index in critically ill patients with established ASCVD remains underexplored. Our study addresses this gap by analyzing a large ASCVD cohort with robust methodological approaches, including multivariable adjustment, propensity score matching, and mediation analysis. These strengths enhance the reliability of our findings and support TyG as a practical biomarker for early risk stratification in this high-risk population.
The association between TyG and mortality appeared more pronounced in nondiabetic patients. This may reflect acute, stress-induced insulin resistance triggered by critical illness in the absence of chronic metabolic adaptation, making nondiabetic individuals more vulnerable to metabolic and inflammatory dysregulation [26, 27]. Such stress-induced IR is often accompanied by heightened inflammatory and oxidative stress responses, potentially accelerating disease progression. In contrast, diabetic patients often receive glucose-lowering therapies such as insulin or SGLT2 inhibitors, which may reduce systemic inflammation and oxidative damage, thereby attenuating the impact of elevated TyG [28, 29]. Additionally, the higher burden of comorbidities among diabetics may dilute the incremental prognostic effect of TyG. These findings suggest that nondiabetic status could modify the association between TyG and outcomes, consistent with prior studies reporting stronger associations in nondiabetic population [11, 30–32]. However, the subgroup analyses were conducted post hoc, without pre-specified sample size or power calculations, and the relatively small strata may reduce the robustness of these observations. Therefore, the results should be regarded as exploratory and interpreted with caution. Future adequately powered studies with pre-specified subgroup analyses are warranted to validate whether stress-induced insulin resistance and its inflammatory consequences differ between diabetic and nondiabetic critically ill patients.
Mediation analysis suggested that elevated WBC count may partially mediate the association between the TyG index and mortality. These findings are exploratory and should be interpreted with caution [33–36]. As a surrogate marker of IR, an elevated TyG index reflects dysregulation in glucose and lipid metabolism, which can activate proinflammatory pathways (e.g., NF-κB, JNK) and promote leukocyte activation [37]. Concurrently, elevated free fatty acids (FFAs) in IR may induce mitochondrial dysfunction and reactive oxygen species (ROS) production, sustaining inflammation and oxidative stress [37]. Several caveats should be noted. First, both TyG and WBC were measured within 24 h of ICU admission, precluding determination of temporal sequence. It remains uncertain whether elevated TyG drives inflammation, whether systemic inflammation drives both higher WBC and altered metabolism, or whether a bidirectional relationship exists. Second, WBC is a nonspecific marker influenced by infection, stress, or steroid therapy, and cannot fully capture the complexity of the insulin resistance–inflammation axis. Other unmeasured mediators—including CRP, IL-6, TNF-α, markers of endothelial dysfunction, and prothrombotic activity—may also contribute but were unavailable in the MIMIC-IV database. Taken together, the mediation findings should be viewed as exploratory rather than confirmatory. Prospective studies with longitudinal profiling of specific inflammatory and endothelial biomarkers are needed to clarify the mechanistic pathways linking insulin resistance, systemic inflammation, and mortality in critically ill ASCVD patients. In addition, patient characteristics such as age, body weight, comorbidities, and illness severity may also influence both WBC levels and mortality risk, and thus could confound the observed mediation effect.
Collectively, our findings support the TyG index as a prognostic biomarker and a potential tool for early risk stratification in critical illness, particularly in patients at risk of rapid deterioration who may require early initiation of MV or vasopressor support. The clinical utility of these findings is particularly relevant among high-risk ASCVD populations, where metabolic dysregulation is prevalent: integrating TyG assessment into routine ICU evaluation may facilitate timely recognition of clinical decline and inform proactive management strategies. Therapies targeting metabolic pathways, such as lipid-lowering or insulin-sensitizing agents, may be beneficial. Although both lifestyle and pharmacological approaches that are aimed at reducing TyG appear promising, prospective studies are needed to determine whether lowering TyG can translate into improved clinical outcomes and help inform metabolically tailored care strategies in critical care settings.
In addition to the primary findings, certain methodological aspects of the secondary outcomes warrant clarification. We used linear regression to estimate the association between TyG index and ventilator- or vasopressor-free days within 28 days, as these variables approximated normal distribution and residuals met model assumptions. Nevertheless, linear regression does not account for competing events such as death or ICU discharge, which are highly relevant in critically ill populations. To address this limitation, we complemented the analysis with Fine–Gray competing risk models. The consistency of findings across both methods supports the robustness of our results.
This study has several limitations. First, this was a single-center retrospective analysis using data from the MIMIC-IV database. Although the dataset is comprehensive, it primarily includes ICU admissions from a single care center, which may limit the generalizability of our findings to broader or less severely ill ASCVD populations. Second, although we performed multivariable adjustments, propensity score matching, and E-value analysis to mitigate confounding, the possibility of residual confounding from unmeasured variables cannot be excluded. Third, our analyses were based solely on baseline TyG levels at ICU admission; dynamic changes in TyG or related metabolic markers during the ICU stay were not captured. This may underestimate or overestimate the true association between TyG and outcomes. Future studies incorporating longitudinal TyG trajectories may provide additional prognostic insight. Fourth, more than 10,000 patients were excluded due to missing triglyceride or fasting glucose values. Post hoc comparisons showed that excluded patients were generally older, had more comorbidities, and worse laboratory profiles, suggesting that the analytic cohort may represent a relatively healthier subgroup. This selection bias may restrict the external validity of our results. Fifth, while mediation analysis suggested systemic inflammation as a partial explanatory pathway, both TyG and WBC were measured within the first 24 h of ICU admission, precluding determination of temporal sequence; moreover, causality cannot be definitively inferred due to the observational design. Despite these limitations, our findings offer robust and novel evidence supporting the prognostic and mechanistic relevance of TyG in critically ill ASCVD patients and lay the groundwork for future prospective, mechanistic, and interventional studies.
Conclusion
In summary, this study found that the TyG index was independently associated with mortality in critically ill ASCVD patients. As a simple, scalable biomarker, TyG holds promise as a potential tool for early risk stratification and guiding metabolic-inflammatory interventions in high-risk cardiovascular populations. These findings underscore the importance of integrating metabolic parameters into critical care management to improve outcomes in ASCVD.
Supplementary Information
Supplementary material 1: Figure 1 E-value bias plot for the association between the highest TyG tertile and 30-day mortality. The solid curve shows the minimum strength of association, on the risk ratio scale, that an unmeasured confounder would need to have with both the exposure (TyG index) and the outcome (30-day mortality) to fully explain away the observed hazard ratio of 1.79. The dashed line corresponds to the lower bound of the 95% confidence interval (HR = 1.56). The plot demonstrates that considerable unmeasured confounding would be required to negate the observed association.
Supplementary material 2: Figure 2 Forest plots of hazard ratios (HRs) for 30-, 90-, and 365-day mortality across TyG tertiles after propensity score matching (PSM). Patients were matched 1:1 across tertiles (T1 = 251, T2 = 251, T3 = 251). Cox proportional hazards models showed that compared with the lowest tertile (T1), the highest tertile (T3) was significantly associated with increased risk of mortality at all time points, whereas T2 vs. T1 was not significant. Results are presented as HRs with 95% confidence intervals (CIs).
Supplementary material 3: Table 1 Disease codes included in the study.
Supplementary material 4: Table 2 Sample Missing Data Overview.
Supplementary material 5: Table 3 Variance inflation factors (VIFs) for covariates included in the multivariable regression models.
Supplementary material 6: Table 4 Comparison of Baseline Characteristics, Clinical Treatments, and Prognostic Indicators Between Patients With and Without TyG Index.
Supplementary material 7: Table 5 Baseline characteristics and outcomes of participants classified by TyG index tertiles after Propensity Score Matching.
Supplementary material 8: Table 6 Logistic Regression Analyses of the Association Between TyG Index and First-Day Mechanical Ventilation and Vasopressor Use.
Supplementary material 9: Table 7 Linear Regression Analysis of the Association Between TyG Index and Ventilator- and Vasopressor-Free Days at Day 28.
Supplementary material 10: Table 8 Univariate and Multivariate Cox Regression Analyses of the Association Between TyG Index and 30-, 90-, and 365-Day Mortality in Patients With ASCVD After Excluding Patients with Malignancy.
Supplementary material 11: Table 9 Univariate and Multivariate Cox Regression Analyses of the Association Between TyG Index and 30-, 90-, and 365-Day Mortality in Patients with ASCVD After Excluding Patients with Malignancy, Liver Failure, and End-Stage Renal Disease.
Acknowledgements
We thank Liu Jie (People’s Liberation Army of China General Hospital, Beijing, China) for helping with statistical support.
Abbreviations
- CVD
Cardiovascular disease
- ASCVD
Atherosclerotic cardiovascular disease
- IR
Insulin resistance
- TYG
Triglyceride glucose index
- MV
Mechanical ventilation
- MIMIC-IV
Medical Information Mart for Intensive Care
- STROBE
Strengthening the Reporting of Observational Studies in Epidemiology
- VIFs
Variance inflation factors
- SD
Standard deviation
- IQR
Interquartile range
- RCS
Restricted cubic spline
- HR
Hazard ratio
- CI
Confidence interval
Author contributions
Huibo Wang designed the study and wrote the main manuscript text. Guosong Jiang conducted the statistical analysis and data interpretation, and performed relevant literature retrieval. Xiaoxiao Qu contributed to the interpretation of the results and reviewed the manuscript. Jinmeng Zhou supervised the research, provided guidance on the study design, critically revised the manuscript, and provided final approval for submission. All authors reviewed and edited the report and have seen and approved the final draft.
Funding
No funding was received for this study.
Data availability
The datasets supporting the results of the present study are available in the MIMIC-IV database (https://physionet.org/content/mimiciv/3.1/).
Declarations
Ethics approval and consent to participate
The Institutional Review Boards (IRBs) of the MIT and BIDMC granted ethical clearance for retrospective analysis of the de-identified dataset. Considering the retrospective nature and anonymized characteristics of the dataset, approval for waiving informed consent was obtained. All authors agreed to the publication of the article.
Consent for publication
Not applicable.
Competing interests
The authors declare no competing interests.
Footnotes
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- 1.Roth GA, Mensah GA, Johnson CO, Addolorato G, Ammirati E, Baddour LM, et al. Global burden of cardiovascular diseases and risk factors, 1990–2019: update from the GBD 2019 study. J Am Coll Cardiol. 2020;76:2982–3021. 10.1016/j.jacc.2020.11.010. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Mensah GA, Fuster V, Murray CJL, Roth GA. Global burden of cardiovascular diseases and risks, 1990-2022. J Am Coll Cardiol. 2023;82:2350–473. 10.1016/j.jacc.2023.11.007. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Wu G, Yu G, Zheng M, Peng W, Li L. Recent advances for dynamic-based therapy of atherosclerosis. Int J Nanomedicine. 2023;18:3851–78. 10.2147/ijn.S402678. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Yi J, Qu C, Li X, Gao H. Insulin resistance assessed by estimated glucose disposal rate and risk of atherosclerotic cardiovascular diseases incidence: the multi-ethnic study of atherosclerosis. Cardiovasc Diabetol. 2024;23:349. 10.1186/s12933-024-02437-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Wang F, Guo Y, Tang Y, Zhao S, Xuan K, Mao Z, et al. Combined assessment of stress hyperglycemia ratio and glycemic variability to predict all-cause mortality in critically ill patients with atherosclerotic cardiovascular diseases across different glucose metabolic states: an observational cohort study with machine learning. Cardiovasc Diabetol. 2025;24:199. 10.1186/s12933-025-02762-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Brunham LR, Lonn E, Mehta SR. Dyslipidemia and the current state of cardiovascular disease: epidemiology, risk factors, and effect of lipid lowering. Can J Cardiol. 2024;40:S4-s12. 10.1016/j.cjca.2024.04.017. [DOI] [PubMed] [Google Scholar]
- 7.Virani SS, Alonso A, Aparicio HJ, Benjamin EJ, Bittencourt MS, Callaway CW, et al. Heart disease and stroke statistics-2021 update: a report from the American Heart Association. Circulation. 2021;143:e254–743. 10.1161/cir.0000000000000950. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Lebovitz HE. Insulin resistance: definition and consequences. Exp Clin Endocrinol Diabetes. 2001;109:S135–48. 10.1055/s-2001-18576. [DOI] [PubMed] [Google Scholar]
- 9.Artunc F, Schleicher E, Weigert C, Fritsche A, Stefan N, Häring HU. The impact of insulin resistance on the kidney and vasculature. Nat Rev Nephrol. 2016;12:721–37. 10.1038/nrneph.2016.145. [DOI] [PubMed] [Google Scholar]
- 10.Simental-Mendía LE, Rodríguez-Morán M, Guerrero-Romero F. The product of fasting glucose and triglycerides as surrogate for identifying insulin resistance in apparently healthy subjects. Metab Syndr Relat Disord. 2008;6:299–304. 10.1089/met.2008.0034. [DOI] [PubMed] [Google Scholar]
- 11.Zhang R, Shi S, Chen W, Wang Y, Lin X, Zhao Y, et al. Independent effects of the triglyceride-glucose index on all-cause mortality in critically ill patients with coronary heart disease: analysis of the MIMIC-III database. Cardiovasc Diabetol. 2023;22:10. 10.1186/s12933-023-01737-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Chen M, Yang Y, Hu W, Gong L, Liao Z, Fu Y, et al. Association between triglyceride-glucose index and prognosis in critically ill patients with acute coronary syndrome: evidence from the MIMIC database. Int J Med Sci. 2025;22:1528–41. 10.7150/ijms.107976. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Cai W, Xu J, Wu X, Chen Z, Zeng L, Song X, et al. Association between triglyceride-glucose index and all-cause mortality in critically ill patients with ischemic stroke: analysis of the MIMIC-IV database. Cardiovasc Diabetol. 2023;22:138. 10.1186/s12933-023-01864-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Wang H, Fu Q, Xiao S, Ma X, Liao Y, Kang C, et al. Predictive value of the triglyceride-glucose index for short- and long-term all-cause mortality in patients with critical coronary artery disease: a cohort study from the MIMIC-IV database. Lipids Health Dis. 2024. 10.1186/s12944-024-02252-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Johnson AEW, Bulgarelli L, Shen L, Gayles A, Shammout A, Horng S, et al. MIMIC-IV, a freely accessible electronic health record dataset. Sci Data. 2023. 10.1038/s41597-022-01899-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Yang Q, Zheng J, Chen W, Chen X, Wen D, Chen W, et al. Association between preadmission Metformin use and outcomes in intensive care unit patients with sepsis and type 2 diabetes: a cohort study. Front Med (Lausanne). 2021. 10.3389/fmed.2021.640785. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP. The strengthening the reporting of observational studies in epidemiology (STROBE) statement: guidelines for reporting observational studies. Lancet. 2007;370:1453–7. 10.1016/s0140-6736(07)61602-x. [DOI] [PubMed] [Google Scholar]
- 18.Austin PC, White IR, Lee DS, van Buuren S. Missing data in clinical research: a tutorial on multiple imputation. Can J Cardiol. 2021;37:1322–31. 10.1016/j.cjca.2020.11.010. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Lv L, Xiong J, Huang Y, He T, Zhao J. Association between the triglyceride glucose index and all-cause mortality in critically ill patients with acute kidney injury. Kidney Dis (Basel, Switzerland). 2024;10:69–78. 10.1159/000535891. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Zheng R, Qian S, Shi Y, Lou C, Xu H, Pan J. Association between triglyceride-glucose index and in-hospital mortality in critically ill patients with sepsis: analysis of the MIMIC-IV database. Cardiovasc Diabetol. 2023;22:307. 10.1186/s12933-023-02041-w. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Huang Y, Li Z, Yin X. Triglyceride-glucose index: a novel evaluation tool for all-cause mortality in critically ill hemorrhagic stroke patients-a retrospective analysis of the MIMIC-IV database. Cardiovasc Diabetol. 2024;23:100. 10.1186/s12933-024-02193-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Zhang F, Hou X. Association between the triglyceride glucose index and heart failure: NHANES 2007-2018. Front Endocrinol (Lausanne). 2023;14:1322445. 10.3389/fendo.2023.1322445. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Zheng D, Cao L. Association between myocardial infarction and triglyceride-glucose index: a study based on NHANES database. Glob Heart. 2024;19:23. 10.5334/gh.1303. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Liu C, Liang D, Xiao K, Xie L. Association between the triglyceride-glucose index and all-cause and CVD mortality in the young population with diabetes. Cardiovasc Diabetol. 2024;23:171. 10.1186/s12933-024-02269-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Liu Q, Zhang Y, Chen S, Xiang H, Ouyang J, Liu H, et al. Association of the triglyceride-glucose index with all-cause and cardiovascular mortality in patients with cardiometabolic syndrome: a national cohort study. Cardiovasc Diabetol. 2024;23:80. 10.1186/s12933-024-02152-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Chen S, Mei Q, Guo L, Yang X, Luo W, Qu X, et al. Association between triglyceride-glucose index and atrial fibrillation: a retrospective observational study. Front Endocrinol. 2022;13:1047927. 10.3389/fendo.2022.1047927. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Laakso M, Kuusisto J. Insulin resistance and hyperglycaemia in cardiovascular disease development. Nat Rev Endocrinol. 2014;10:293–302. 10.1038/nrendo.2014.29. [DOI] [PubMed] [Google Scholar]
- 28.Xu M, Zheng J, Hou T, Lin H, Wang T, Wang S, et al. SGLT2 inhibition, choline metabolites, and cardiometabolic diseases: a mediation Mendelian randomization study. Diabetes Care. 2022;45:2718–28. 10.2337/dc22-0323. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Kim SR, Lee SG, Kim SH, Kim JH, Choi E, Cho W, et al. SGLT2 inhibition modulates NLRP3 inflammasome activity via ketones and insulin in diabetes with cardiovascular disease. Nat Commun. 2020;11:2127. 10.1038/s41467-020-15983-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.van den Berghe G, Wouters P, Weekers F, Verwaest C, Bruyninckx F, Schetz M, et al. Intensive insulin therapy in critically ill patients. N Engl J Med. 2001;345:1359–67. 10.1056/NEJMoa011300. [DOI] [PubMed] [Google Scholar]
- 31.Agrawal NK, Kant S. Targeting inflammation in diabetes: newer therapeutic options. World J Diabetes. 2014;5:697–710. 10.4239/wjd.v5.i5.697. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Liao Y, Zhang R, Shi S, Zhao Y, He Y, Liao L, et al. Triglyceride-glucose index linked to all-cause mortality in critically ill patients: a cohort of 3026 patients. Cardiovasc Diabetol. 2022;21:128. 10.1186/s12933-022-01563-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Rohm TV, Meier DT, Olefsky JM, Donath MY. Inflammation in obesity, diabetes, and related disorders. Immunity. 2022;55:31–55. 10.1016/j.immuni.2021.12.013. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Ormazabal V, Nair S, Elfeky O, Aguayo C, Salomon C, Zuñiga FA. Association between insulin resistance and the development of cardiovascular disease. Cardiovasc Diabetol. 2018;17:122. 10.1186/s12933-018-0762-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Vlachopoulos C, Dima I, Aznaouridis K, Vasiliadou C, Ioakeimidis N, Aggeli C, et al. Acute systemic inflammation increases arterial stiffness and decreases wave reflections in healthy individuals. Circulation. 2005;112:2193–200. 10.1161/circulationaha.105.535435. [DOI] [PubMed] [Google Scholar]
- 36.Reaven GM. The insulin resistance syndrome: definition and dietary approaches to treatment. Annu Rev Nutr. 2005;25:391–406. 10.1146/annurev.nutr.24.012003.132155. [DOI] [PubMed] [Google Scholar]
- 37.Gremmels H, Bevers LM, Fledderus JO, Braam B, van Zonneveld AJ, Verhaar MC, et al. Oleic acid increases mitochondrial reactive oxygen species production and decreases endothelial nitric oxide synthase activity in cultured endothelial cells. Eur J Pharmacol. 2015;751:67–72. 10.1016/j.ejphar.2015.01.005. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Supplementary material 1: Figure 1 E-value bias plot for the association between the highest TyG tertile and 30-day mortality. The solid curve shows the minimum strength of association, on the risk ratio scale, that an unmeasured confounder would need to have with both the exposure (TyG index) and the outcome (30-day mortality) to fully explain away the observed hazard ratio of 1.79. The dashed line corresponds to the lower bound of the 95% confidence interval (HR = 1.56). The plot demonstrates that considerable unmeasured confounding would be required to negate the observed association.
Supplementary material 2: Figure 2 Forest plots of hazard ratios (HRs) for 30-, 90-, and 365-day mortality across TyG tertiles after propensity score matching (PSM). Patients were matched 1:1 across tertiles (T1 = 251, T2 = 251, T3 = 251). Cox proportional hazards models showed that compared with the lowest tertile (T1), the highest tertile (T3) was significantly associated with increased risk of mortality at all time points, whereas T2 vs. T1 was not significant. Results are presented as HRs with 95% confidence intervals (CIs).
Supplementary material 3: Table 1 Disease codes included in the study.
Supplementary material 4: Table 2 Sample Missing Data Overview.
Supplementary material 5: Table 3 Variance inflation factors (VIFs) for covariates included in the multivariable regression models.
Supplementary material 6: Table 4 Comparison of Baseline Characteristics, Clinical Treatments, and Prognostic Indicators Between Patients With and Without TyG Index.
Supplementary material 7: Table 5 Baseline characteristics and outcomes of participants classified by TyG index tertiles after Propensity Score Matching.
Supplementary material 8: Table 6 Logistic Regression Analyses of the Association Between TyG Index and First-Day Mechanical Ventilation and Vasopressor Use.
Supplementary material 9: Table 7 Linear Regression Analysis of the Association Between TyG Index and Ventilator- and Vasopressor-Free Days at Day 28.
Supplementary material 10: Table 8 Univariate and Multivariate Cox Regression Analyses of the Association Between TyG Index and 30-, 90-, and 365-Day Mortality in Patients With ASCVD After Excluding Patients with Malignancy.
Supplementary material 11: Table 9 Univariate and Multivariate Cox Regression Analyses of the Association Between TyG Index and 30-, 90-, and 365-Day Mortality in Patients with ASCVD After Excluding Patients with Malignancy, Liver Failure, and End-Stage Renal Disease.
Data Availability Statement
The datasets supporting the results of the present study are available in the MIMIC-IV database (https://physionet.org/content/mimiciv/3.1/).