Abstract
PURPOSE: To explore how diabetes influences the association between stress hyperglycemia (SH) and functional outcomes in acute ischemic stroke (AIS), and to develop a nomogram based on the stress hyperglycemia ratio (SHR) to predict the prognosis of AIS patients. METHODS: Patients were divided into diabetic mellitus (DM) and non-diabetic mellitus (non-DM) groups. SH was calculated using the ratio of fasting blood sugar (FBG) to glycosylated hemoglobin A1c (HbA1c). Poor functional outcomes were defined as a modified Rankin Scale (mRS) score > 2 at a 3-month follow-up. Multivariable logistic regression analysis was used to identify the relationship between SHR and functional outcomes in diabetic and non-diabetic groups. Least Absolute Shrinkage Selection Operator (LASSO) regression analysis and logistic regression analysis were employed to build a nomogram. RESULTS: In the DM group, severe stress hyperglycemia showed no significant association with poor function outcomes at 3 months. However, in the non-DM group, severe stress hyperglycemia was associated with poor function outcomes [odds ratio (OR) = 2.21, 95% confidence interval CI 1.61-3.02, p < 0.001]. After adjusting for covariates, the differences remained statistically significant (adjusted OR = 1.62, 95% CI 1.09-2.42, p = 0.0181). Lasso regression identified 8 predictive factors, while logistic regression highlighted age, NIHSS, and SHR as independent predictors, forming a nomogram. The receiver operating characteristic curve assessed the discriminative ability of the nomogram, calibration, and clinical decision curves to evaluate model fitting and net benefit. CONCLUSION: Severe stress hyperglycemia is linked to worse outcomes in non-diabetic patients but not significantly correlated in diabetic patients. A nomogram incorporating age, SHR, and NIHSS predicts poor outcomes in non-diabetic patients.