Evaluation of four insulin resistance surrogate indicators for predicting short-term mortality in critically ill patients with ischemic stroke: a nationwide retrospective cohort study

BACKGROUND: Insulin resistance (IR) is closely linked to the incidence and adverse outcomes of acute ischemic stroke (IS). However, the prognostic value of various surrogate IR indices remains inconsistent across studies. This study aimed to compare the associations and discriminative abilities of four surrogate IR indices-the triglyceride-glucose index (TyG), TyG-body mass index (TyG-BMI), atherogenic index of plasma (AIP), and metabolic score for insulin resistance (METS-IR)-in relation to 28-day in-hospital mortality among critically ill patients with IS. METHODS: This study utilized data from the eICU database to investigate critically ill patients with IS. Associations between the four IR indices and short-term mortality were assessed using multivariate regression and Cox regression combined with restricted cubic spline modeling (RCS). Additionally, receiver operating characteristic (ROC) analysis was conducted to evaluate the discriminative performance of these surrogate IR indices. Stratified analyses were performed to identify potential interactions among demographic variables. RESULTS: A total of 1346 critically ill patients with IS (52.60% female) were enrolled in the study. The overall 28-day mortality rate was 8.77%, with no significant differences between genders. Multivariate regression analysis indicated that both the TyG and the METS.IR were independently associated with increased mortality, with Hazard Ratios (HR) of 1.44 and 1.41, respectively. Dose-response analyses revealed linear associations of TyG and AIP with short-term mortality risk, whereas TyG-BMI and METS-IR exhibited nonlinear relationships with mortality risk that plateaued at higher values. ROC analysis demonstrated that the TyG index had relatively superior discriminative performance among the four IR surrogate indicators in the overall population, male patients, and those over 60 years of age. In contrast, AIP (AUC = 0.68) showed stronger discriminative ability in patients under 60 years compared to the other indices. Subgroup analyses further confirmed significant associations of TyG with mortality in males (HR = 1.49, 95% CI 1.17-1.90; p = 0.0013) and of AIP in patients under 60 years (HR = 1.58, 95% CI 1.07-2.32; p = 0.0204). No significant interactions were observed across sex or age subgroups for the four IR surrogate indices. CONCLUSIONS: Four surrogate indices of IR were positively associated with short-term mortality risk in critically ill patients with IS. Among these indices, TyG and METS-IR demonstrated relatively stronger risk effects, while the AIP exhibited superior effect size and discriminative ability in patients under 60 years of age. In contrast, TyG-BMI displayed comparatively weaker performance both in the overall population and across subgroups. These findings address a critical gap in understanding the comparative utility of surrogate IR indices for assessing short-term mortality risk in severe IS. Routine evaluation of these indices may enhance clinical assessment and inform individualized patient management across different subgroups. Furthermore, incorporating robust indicators such as TyG into clinical risk assessment models could enhance the precision of mortality predictions and facilitate optimized patient management strategies.