Predictive value of admission blood glucose level on short-term mortality in acute cerebral ischemia

Background: Admission hyperglycemia increases the risk of death in patients with acute stroke. However, the most appropriate cut-off of glucose level indicating an increased risk of short-term mortality remains unknown. Purpose and methods: We aimed at establishing the optimum cut-offs of several variables (including admission blood glucose levels) predicting case-fatality (72 hours, 7days) and unfavorable outcome [modified Rankin Scale (mRS) score 5-6 at 7 days] in consecutive first-ever acute ischemic stroke. Receiver operating characteristic (ROC) curves were constructed. Results: Eight hundred eleven consecutive patients were included [median age of 77 (69-83) years; 418 (52%) male; 239 (30%) diabetics; median admission National Institutes of Health Stroke Scale (NIHSS) 7 (4-12), 32 (4%) dead within 72 hours; 64 (8%) dead within day 7: 155 (19%) with unfavorable outcome]. Median admission glucose levels were 113 (97-155) mg/dL. Diabetics had significantly higher median glucose levels than non-diabetics [163 (133-214) vs. 107 (92-123) mg/dL, p<0.001]. According to ROC analysis, the only significant predictive value of glycemia was >= 143 mg/dL for 72-hour fatality (sensitivity 88% and specificity 70%) especially in non-diabetics (sensitivity 88% and sensitivity 62%). This cut-off point was an independent predictor for 72-hour fatality (overall: OR = 4.0, CI = 1.6-9.9, p = 0.003; non-diabetics: OR = 4.9, CI = 1.7-14.5, p = 0.004). The cut-offs of fasting total cholesterol levels and admission leukocytes had poor predictive values for each outcome, while those of admission NIHSS had good discrimination in predicting short-term outcome measures. Conclusions: Admission hyperglycemia (>= 143 mg/dL) is a strong and an independent predictor for 72-hour fatality, especially in patients with no prior history of diabetes mellitus. (C) 2012 Elsevier Inc. All rights reserved.