Comparison of predictive scores of symptomatic intracerebral haemorrhage after stroke thrombolysis in a single centre.

UNLABELLED Symptomatic intracerebral haemorrhage following thrombolysis for ischaemic stroke causes significant morbidity and mortality. This study assessed which of four risk scores (SEDAN, HAT, GRASPS and SITS) best predicts symptomatic intracerebral haemorrhage. METHODS Data from 431 patients treated at Aberdeen Royal Infirmary (2003-2013) were extracted from a thrombolysis database. Score performance was compared using area under the curve. RESULTS Any intracerebral haemorrhage occurred in 12% of patients (53/413); 11% fulfilling the SITS-MOST symptomatic intracerebral haemorrhage definition (6/53), 34% the ECASS II definition (18/53), and 43% the National Institute of Neurological Disorder and Stroke definition (23/53). Stroke severity, as defined by the National Institutes of Health Stroke Scale, significantly improved after 24 hours in patients without intracerebral haemorrhage, but not in those with. Significant symptomatic intracerebral haemorrhage predictors were age, glucose, stroke severity, hyperdense middle cerebral artery on CT scan, ASPECTS score and anti-platelet therapy. The haemorrhage after thrombolysis score performed best at predicting symptomatic intracerebral haemorrhage (area under the curve 0.67-0.78, p < 0.001). CONCLUSION The haemorrhage after thrombolysis score uses the least variables and has the best predictive value for symptomatic intracerebral haemorrhage. Using predictive scores for clinical decision making depends on estimation of overall benefits as well as risk.