# Prognostic value of the triglyceride–glucose index for ICU mortality in non-diabetic sepsis: a restricted cubic spline analysis

## TL;DR

The triglyceride–glucose (TyG) index is linked to higher ICU mortality in non-diabetic sepsis patients, with a U-shaped pattern suggesting it could help predict risk.

## Contribution

This study identifies a U-shaped relationship between TyG index and ICU mortality in non-diabetic sepsis patients, validated in an external cohort.

## Key findings

- ICU mortality was 12.2% in the MIMIC cohort, with higher risk in the highest TyG quartile.
- RCS analysis showed a U-shaped association with a threshold at TyG = 9.163.
- External validation showed higher mortality in the high-TyG group, though not statistically significant.

## Abstract

The triglyceride–glucose (TyG) index has been identified as a metabolic marker associated with adverse outcomes in sepsis, but its prognostic value in non-diabetic septic patients remains unclear.

To assess the association between TyG levels and ICU mortality in non-diabetic sepsis and validate the findings in an external cohort.

A retrospective analysis of 2,217 non-diabetic sepsis patients from the MIMIC-IV database was conducted using multivariate logistic regression, threshold effect analysis, restricted cubic spline (RCS) modeling, and subgroup analyses. External validation was performed in an independent ICU cohort of 185 non-diabetic sepsis patients from The Central Hospital of Enshi, stratified by the MIMIC-derived TyG cut-off of 9.163.

ICU mortality in the MIMIC cohort was 12.2%. Compared with Q2, Q4 showed a significantly increased mortality risk (OR = 1.49, 95% CI: 1.12–1.97, P = 0.006). RCS analysis demonstrated a significant U-shaped association with a threshold at TyG = 9.163. Subgroup analyses confirmed consistent trends. In the external validation cohort, mortality was higher in the high-TyG group than in the low-TyG group (29.0% vs. 21.7%), showing a directionally consistent, though non-significant, trend (P = 0.23).

The TyG index is independently associated with ICU mortality in non-diabetic sepsis and exhibits a clear U-shaped pattern. The external validation cohort demonstrated a similar risk trend, supporting the broader applicability of TyG as a simple metabolic marker for risk stratification.