부산대학교 도서관

BACKGROUND: AIMS: METHODS: RESULTS: The rate of partial or complete recanalization (thrombolysis in myocardial infarction 2–3) did not differ between patients with and without diabetes (67% vs. 66%; P = 1·000). Mean admission glucose values were similar in patients with poor recanalization (thrombolysis in myocardial infarction 0–1) and patients with partial or complete recanalization (thrombolysis in myocardial infarction 2–3; 7·3 vs. 7·3 mmol/l; P = 0·746). Follow-up at three-months was obtained in 388 of 389 patients. Clinical outcome was favourable (modified Rankin Scale 0–2) in 189 patients (49%) and poor (modified Rankin Scale 3–6) in 199 patients (51%). Mortality at three-months was 20%. Diabetics were more likely to have poor outcome (72% vs. 48%; P = 0·001) and to be dead (30% vs. 19%; P = 0·044) at three-months. After multivariable analysis, there remained an independent relationship between diabetes and outcome (P = 0·003; odds ratio 3·033, 95% confidence interval 1·452–6·336), but not with mortality (P = 0·310; odds ratio 1·436; 95% confidence interval 0·714–2·888). Moreover, higher age (P = 0·001; odds ratio 1·039; 95% confidence interval 1·017–1·061), higher baseline National Institutes of Health Stroke Scale score (P < 0·0001; odds ratio 1·130; 95% confidence interval 1·079–1·182), location of vessel occlusion as categorical variable (P < 0·0001), poor collaterals (P = 0·02; odds ratio 1·587; 95% confidence interval 1·076–2·341), poor vessel recanalization (P < 0·0001; odds ratio 4·713; 95% confidence interval 2·627–8·454), and higher leucocyte count (P = 0·032; odds ratio 1·094; 95% confidence interval 1·008–1·188) were independent baseline predictors of poor outcome. Higher admission glucose was associated with poor outcome (P = 0·006) and mortality (P < 0·0001). After multivariate analyses, glucose remained independently associated with poor outcome (P = 0·019; odds ratio 1·150; 95% confidence interval 1·023–1–292) and mortality (P = 0·005; odds ratio 1·183; 95% confidence interval 1052–1·331). The rate of symptomatic intracranial haemorrhage was similar in diabetics and non-diabetics (6·7% vs. 4·6%; P = 0·512). Mean admission glucose was higher in patients with symptomatic intracranial haemorrhage than without (8·58 vs. 7·26 mmol/l; P = 0·010). Multivariable analysis confirmed an independent association between admission glucose and symptomatic intracranial haemorrhage (P = 0·027; odds ratio 1·187; 95% confidence interval 1·020–1·381). CONCLUSIONS