Glycemic Gap Predicts in-Hospital Mortality in Diabetic Patients with Intracerebral Hemorrhage.

BACKGROUND AND PURPOSE

The relationship between admission hyperglycemia and intracerebral hemorrhage (ICH) outcome remains controversial. Glycemic gap (GG) is a superior indicator of glucose homeostatic response to physical stress compared to admission glucose levels. We aimed to evaluate the association between GG and in-hospital mortality in ICH.

METHODS

We retrospectively identified consecutive patients hospitalized for spontaneous ICH at the 2 healthcare systems in the Twin Cities area, MN, between January 2008 and December 2017. Patients without glycosylated hemoglobin (HbA1c) test or those admitted beyond 24 hours post-ICH were excluded. Demographics, medical history, admission tests, and computed tomography data were recorded. GG was computed using admission glucose level minus HbA1c-derived average glucose. The association between GG and time to in-hospital mortality was evaluated by Cox regression analysis. Receiver operating characteristic (ROC) analysis with the DeLong test was used to evaluate the ability of GG to predict in-hospital death.

RESULTS

Among 345 included subjects, 63 (25.7%) died during the hospital stay. Compared with survivors, non-survivors presented with a lower Glasgow coma scale score, larger hematoma volume, and higher white blood cells count, glucose, and GG levels at admission (p<0.001). GG remained an independent predictor of in-hospital mortality after adjusting for known ICH outcome predictors and potential confounders [adjusted hazard ratio: 1.09, 95% confidence interval (CI): 1.02-1.18, p = 0.018]. GG showed a good discriminative power (area under the ROC curve: 0.75, 95% CI: 0.68-0.82) in predicting in-hospital death and performed better than admission glucose levels in diabetic patients (p = 0.030 for DeLong test).

CONCLUSIONS

Admission GG is associated with the risk of in-hospital mortality and can potentially represent a useful prognostic biomarker for ICH patients with diabetes.