Validation assessment of risk scores to predict postthrombolysis intracerebral haemorrhage

Background Two clinical risk scores, the Haemorrhage After Thrombolysis and Multicentre Stroke Survey scores, have been proposed to predict the risk of intracerebral haemorrhage following thrombolysis in acute ischaemic stroke. Aims To validate Haemorrhage After Thrombolysis and Multicentre Stroke Survey scores as predictors of post-tissue plasminogen activator symptomatic intracerebral haemorrhage and asymptomatic intracerebral haemorrhage in an independent cohort. Methods Haemorrhage After Thrombolysis and Multicentre Stroke Survey scores were calculated for the cohort of tissue plasminogen activator-treated patients enrolled in the placebo arms of the SAINT-I and SAINT-II trials. The absolute risk of symptomatic intracerebral haemorrhage and asymptomatic intracerebral haemorrhage associated with each scoring system was determined. The overall predictive value was assessed using c-statistics. Results Symptomatic intracerebral haemorrhage occurred in 5 center dot 6% of 965 patients treated with tissue plasminogen activator in the SAINT cohorts. The risk of symptomatic intracerebral haemorrhage was modestly greater, with higher Haemorrhage After Thrombolysis scores (0: 4 center dot 1%, 1: 4 center dot 1%, 2: 8 center dot 8%, 3: 12 center dot 5%, 4: 0%, 5: no subjects). Similar results were seen with the Multicentre Stroke Survey score (0: 0%, 1: 4 center dot 8%, 2: 2 center dot 3%, 3: 7 center dot 3%, 4: 6 center dot 3%). In logistic regression, the Haemorrhage After Thrombolysis score was associated with the risk of symptomatic intracerebral haemorrhage (odds ratio=1 center dot 41 per point, 95% confidence interval: 1 center dot 05-1 center dot 89, P=0 center dot 021) and asymptomatic intracerebral haemorrhage (odds ratio=1 center dot 59 per point, 95% confidence interval: 1 center dot 33-1 center dot 92, P < 0 center dot 001). The Multicentre Stroke Survey score was modestly associated with the risk of symptomatic intracerebral haemorrhage (odds ratio=1 center dot 43 per point, 95% confidence interval: 0 center dot 95-2 center dot 15, P=0 center dot 084) and asymptomatic intracerebral haemorrhage (odds ratio=1 center dot 63 per point, 95% confidence interval: 1 center dot 27-2 center dot 08, P < 0 center dot 001). The c-statistic was 0 center dot 59 for predicting symptomatic intracerebral haemorrhage and 0 center dot 61 for asymptomatic intracerebral haemorrhage for both the Haemorrhage After Thrombolysis and the Multicentre Stroke Survey scores. Conclusions While both the Haemorrhage After Thrombolysis and Multicentre Stroke Survey scores were associated with a risk of symptomatic intracerebral haemorrhage, discriminatory ability was limited.