Different effect of hyperglycemia on stroke outcome in non-diabetic and diabetic patients--a cohort study.

Objectives: Relationship between hyperglycemia and stroke outcome is unclear, partly due to the small sample size in most studies, and partly due to lack of consensus concerning the cutoff level for hyperglycemia.

Methods: In a cohort study, we investigated whether on-admission hyperglycemia is an independent predictor for 30-day case fatality by analyzing data of 2496 consecutive computed tomography (CT) verified acute ischemic stroke patients (2077 non-diabetic and 419 diabetic) included in the prospective, hospital-based Debrecen Stroke Database. Instead of using an arbitrary cutoff level for hyperglycemia, quartiles of on-admission glucose level were used for Kaplan-Meier survival curves and Cox proportional hazard modeling.

Results: The four quartiles of serum glucose level were in the range as follows: <5.2 mmol/l, 5.201-6.1 mmol/l, 6.101-7.5 mmol/l, and >7.501 mmol/l (n = 664, 618, 597, and 617, respectively). Among all 2496 participants, the adjusted hazard ratios for death increased with each quartile of admission glucose 1.96 [95% confidence interval (CI): 1.07-3.60; P = 0.03], 1.56 (95% CI: 0.83-2.94; P = 0.17), and 3.04 (95% CI: 1.70-5.44; P < 0.0001) for the second, third, and fourth quartiles, respectively). Upon stratification with respect to diabetes, we found similarly high risk for poor outcome among non-diabetic patients, while the risk was considerably lower among diabetic patients.

Discussion: These data suggest that even mild elevation of on-admission glucose levels is an independent predictor of 30-day case fatality. So, we propose that the ideal target blood glucose level is lower for non-diabetic than diabetic patients.