What question did this study set out to answer?

This study aims to examine the relationship between glucose-to-lymphocyte ratio (GLR), ICU delirium, and mortality.

April 27, 2026Open Access

High glucose-to-lymphocyte ratio increases the risk of delirium and all-cause mortality in intensive care unit patients

Key Points

This study aims to examine the relationship between glucose-to-lymphocyte ratio (GLR), ICU delirium, and mortality.
Analyzed data from 16,055 adult patients using the MIMIC-IV database during ICU admission.
Applied multivariable logistic regression and propensity score matching to assess GLR's association with delirium and mortality.
Conducted Cox proportional hazards models and mediation analysis for mortality evaluation.
Patients in the highest GLR quartile had a 2.51-fold higher risk of delirium compared to the lowest quartile (95% CI: 2.12–2.98; P < 0.001).
Survival analysis indicated increased mortality rates at 28-day (HR = 1.17), 90-day (HR = 1.22), and 1-year (HR = 1.22) for high GLR (all P < 0.05).
Mediation analysis showed that delirium partially mediated the relationship between GLR and 28-day mortality (proportion: 15.235%–19.133%).

Abstract

Delirium is a common acute brain dysfunction in the intensive care unit (ICU) that correlates strongly with poor patient outcomes. The glucose-to-lymphocyte ratio (GLR), as a novel marker reflecting metabolic stress and immune status, may represent a distinct biological signal of homeostatic disruption. This study aims to investigate the relationship between GLR, ICU delirium, and mortality. Using the MIMIC-IV database, we included 16,055 adult patients during their first ICU admission. A multivariable logistic regression with a nested modeling strategy (Models A–F) was applied to examine the association between GLR and delirium, adjusting for a broad range of clinical confounders and multiple established severity indices. The robustness of these findings was validated through 1:1 propensity score matching (PSM) and extensive subgroup analyses. Restricted cubic spline (RCS) analysis explored potential non-linear relationships. Cox proportional hazards models and mediation analysis with the bootstrap method were employed to evaluate mortality risks and the mediating role of delirium. Patients in the highest GLR quartile (Q4) had a 2.51-fold higher risk of delirium compared to those in the lowest quartile (Q1) (95% CI: 2.12–2.98; P < 0.001), with a significant dose-response relationship (P for trend < 0.001). In the fully saturated model (Model F), this significant association between GLR and delirium persisted (OR = 2.51, P < 0.001). These findings remained consistent after 1:1 PSM and across all pre-specified subgroups. ROC analysis demonstrated that GLR had superior discriminative performance (AUC = 0.626) compared with blood glucose alone (AUC = 0.587) or lymphocyte count (AUC = 0.603). Survival analysis indicated significantly increased 28-day (HR = 1.17), 90-day (HR = 1.22), and 1-year (HR = 1.22) mortality for high GLR (all P < 0.05). Mediation analysis confirmed that delirium partially mediated the relationship between GLR and 28-day mortality (proportion: 15.235%–19.133%). Elevated GLR is independently associated with increased risks of delirium and mortality in ICU patients. As a composite marker of metabolic-immune dysregulation, GLR provides incremental predictive value beyond established severity scores, offering a robust biological signal for early risk stratification in critical care.

Bookmark

View Full Paper