Question: In patients who have had a stroke, what are the differences in all-cause mortality and functional outcomes for different blood glucose levels and diabetes status?
Data sources: Studies were identified by searching MEDLINE (1966 to December 2000) and Science Citation Index (1980 to December 1999). Bibliographies of relevant articles were reviewed, and experts in the field were contacted to identify unpublished studies.
Study selection: English-language studies were selected if the design was a cohort study, an inception cohort of patients with stroke was used, blood glucose level was measured within 24 hours of admission, ≥ 80% of the patients had follow-up to hospital discharge or to 1 month, and outcomes were reported in relation to blood glucose levels at admission. Exclusion criteria included exclusive reporting on subarachnoid hemorrhage, transient ischemic attack, and nonstroke causes of focal neurologic deficits.
Data extraction: Data were extracted on sample size, type of stroke (ischemic or hemorrhagic), definition of stress hyperglycemia, percentage of patients with stress hyperglycemia, and outcomes. Outcomes included in-hospital or 30-day all-cause mortality and poor functional recovery.
Main results: 31 articles describing 32 cohort studies met the selection criteria. 4 of the 32 studies presented data separately for patients with ischemic or hemorrhagic stroke; 11 studies included only patients with ischemic stroke; and 17 studies either combined data for patients with ischemic or hemorrhagic stroke or did not state the subtype of stroke. In nondiabetic patients with stroke, the risk for in-hospital or 30-day all-cause mortality (short-term mortality) was 3-fold greater in patients with stress hyperglycemia than in those without it (relative risk [RR] 3.07, 95% CI 2.5 to 3.79). In patients with diabetes, stress hyperglycemia was not associated with short-term mortality after stroke (RR 1.3, CI 0.49 to 3.43). Subgroup analysis showed that in nondiabetic patients with ischemic stroke, the risk for short-term mortality was 3-fold greater in patients with a glucose level > 6.1 to 7.0 mmol/L at admission than in those with a glucose level ≤ 6 mmol/L at admission (RR 3.28, CI 2.32 to 4.64). In patients with diabetes who had ischemic stroke, stress hyperglycemia was not associated with short-term mortality (RR 2.00, CI 0.04 to 90). Stress hyperglycemia was also not associated with short-term mortality after hemorrhagic stroke. In 5 studies, the risk for long-term mortality (3 mo to 1 y) was greater in patients with hyperglycemia than in those without it. In patients without diabetes, the risk for poor functional recovery up to 6 months after stroke was greater in patients with stress hyperglycemia than in those without hyperglycemia on admission (RR 1.41, CI 1.16 to 1.73).
Conclusion: In patients without diabetes who have had ischemic stroke, stress hyperglycemia is associated with an increased risk for in-hospital or 30-day all-cause mortality.