Quantitative Analysis of Stress-Induced Hyperglycemia and Intracranial Blood Volumes for Predicting Mortality After Intracerebral Hemorrhage.

Stress-induced hyperglycemia (SIH) is a neuroendocrine response to acute illness. Although SIH has an adverse association with intracerebral hemorrhage (ICH), quantitative measures and determinants of SIH are not well delineated. In the present study, we objectively evaluated SIH using glycemic gap (GG) and identified its radiological and clinical determinants, with a 5-year retrospective review of charts of ICH patients. We calculated GG using the regression equation (GG = AG -28.7 × HbA1c + 46.7) and evaluated whether GG is an independent predictor of mortality using a multivariate regression model. Radiological volumes of different intracranial compartments were determined using image segmentation software. We correlated GG with different clinical and radiological parameters using Pearson correlation coefficient (PCC), Spearman's rank correlation (SRC), and Wilcoxon rank sum test. Then, we calculated the value of GG associated with mortality. Out of 328 patients, 238 (73%) survived hospitalization and 90 (27%) expired. GG was found to be an independent predictor of mortality (r=0.008, p=0.04). Additionally, GG was positively correlated with intraparenchymal hemorrhage (IPH) volume (PCC=0.185, p<0.01) and intraventricular hemorrhage (IVH) volume (PCC=0.233, p<0.01) and negatively correlated with cerebrospinal fluid (CSF) volume (PCC=-0.151, p<0.01) and brain tissue volume (PCC=-0.099, p=0.08). GG was positively correlated with patients' ICH score (SRC=0.377, p<0.01), Glasgow Coma Scale (GCS) (PCC=-0.356, p<0.01), hydrocephalus (p<0.01), and IVH in the third ventricle (p<0.01). The univariate logistic regression model identified 30.0 mg/dl as the value of GG (AUC=0.655, p<0.01) that predicted mortality with 52.2% sensitivity and 75.2% specificity and defined SIH. In conclusion, GG independently predicts mortality in ICH patients and positively correlates with IPH and IVH volumes. However, causality between the two is not established and would require specifically designed studies.