Triglyceride-glucose index: a novel assessment tool for all-cause mortality in critical stroke patients—a retrospective analysis of the eICU-CRD database

BACKGROUND: Stroke, a leading cause of death and disability, includes ischemic (IS), hemorrhagic (HS), and subarachnoid subtypes. While the triglyceride-glucose index (TyG) has been linked to cerebrovascular diseases, studies focus on single subtypes and lack ICU data. The prognostic roles of the atherogenic index of plasma (AIP) and total cholesterol-to-HDL ratio (TC_HDL) in critically ill stroke patients' all-cause mortality (ACM) remain unexplored. This study investigates these indices' associations with ACM in critically ill stroke patients. METHODS: We screened 3247 ICU-admitted stroke patients from the eICU-CRD, stratified by TyG, AIP, and TC_HDL quartiles. Endpoints were in-hospital and ICU mortality. Analyses used Cox regression, Kaplan-Meier curves, and restricted cubic splines. RESULTS: In-hospital and ICU mortality rates were 17.2% and 9.7%, respectively. Multivariate Cox analysis showed elevated TyG correlated with higher ICU mortality [HR 1.528, 95% CI 1.252-1.771, *P* = 0.002] and in-hospital mortality [HR 1.434, 95% CI 1.04-1.619, *P* = 0.013]. Elevated AIP associated with higher ICU [HR 1.342, *P* = 0.021] and in-hospital mortality [HR 1.631, *P* = 0.005]. TC_HDL showed no significant associations. Kaplan-Meier curves confirmed TyG's links to both mortalities and AIP's association with in-hospital mortality. Restricted cubic splines revealed increasing risk with higher TyG. CONCLUSION: TyG shows stronger correlations with short-term ICU and in-hospital mortality in critically ill stroke patients compared to AIP and TC_HDL. Monitoring TyG may aid prognostic assessment and risk stratification, supporting optimized treatment strategies and aligning with prior research on TyG and stroke outcomes.