Association between triglyceride-glucose index trajectories and in-hospital mortality in sepsis: a cohort study based...

Sepsis remains a major challenge in critical care medicine, characterized by high incidence and mortality rates that severely threaten patient prognosis. Insulin resistance (IR) plays a pivotal role in the metabolic disturbances and adverse outcomes associated with sepsis. The triglyceride-glucose (TyG) index, as a readily attainable surrogate diagnostic for IR, has been frequently employed in clinical studies. The relationship between the TyG index's dynamic trajectories and clinical outcomes is yet unknown, though, as prior research has mostly assessed the index at a single time point.

This retrospective study included ICU patients with sepsis, identified according to the Sepsis-3 criteria, from the MIMIC-IV database (2008-2019). Eligible participants were those aged ≥ 18 years, with first ICU admission, at least three venous blood glucose measurements, and at least one triglyceride measurement. The latent class mixed model (LCMM) was applied to classify dynamic trajectories of the TyG index within the first 72 h of ICU stay. LASSO and Boruta algorithms were jointly used for covariate selection. Subgroup and interaction analyses were conducted in addition to multivariable logistic regression to evaluate the relationship between various TyG trajectories and mortality.

A total of 3,555 sepsis patients were included. Trajectory analysis identified five distinct TyG dynamic patterns. Using the "persistently low" group as the reference, the fully adjusted model showed that the "increase-then-decrease" (OR = 2.61, 95% CI: 1.64-4.16), "decrease-then-increase" (OR = 1.46, 95% CI: 1.01-2.13), and "stable moderate" (OR = 1.23, 95% CI: 1.01-1.50) groups had significantly higher risks of in-hospital mortality. Subgroup analyses indicated that these associations were robust across most clinical strata.

The TyG index exhibits substantial dynamic heterogeneity among ICU patients with sepsis. Certain abnormal trajectories (such as "increase-then-decrease", "decrease-then-increase", and "stable moderate") are associated with a markedly increased risk of in-hospital mortality. TyG trajectory analysis may provide a novel tool for risk stratification and individualized management in sepsis patients.