Abstract
BACKGROUND: Symptomatic intracerebral hemorrhage (sICH) after intravenous recombinant tissue plasminogen activator in patients with acute ischemic stroke (AIS) remains a feared yet unpredictable complication. We aimed to develop and validate a new predictive model incorporating clinical variables and noncontrast head computed tomography imaging features to predict sICH in patients with AIS receiving intravenous recombinant tissue plasminogen activator. METHODS AND RESULTS: The predictive model was derived from 808 patients with AIS in the derivation cohort in Southeast China, based on multivariable logistic regression analysis. External validation was conducted in a validation cohort from Central China. Discrimination, calibration, and clinical usefulness of the predictive model were assessed. We observed 32 sICH events among 808 patients with AIS in the derivation cohort, and 21 sICH events out of 612 participants in the validation cohort. The variables in the predictive model included cerebral small vessel disease burden and early infarct signs on head computed tomography scan, atrial fibrillation, age, systolic blood pressure, and initial National Institutes of Health Stroke Scale score. The fitted model showed promising discrimination (optimism-corrected C statistic of 0.80) and acceptable calibration (Hosmer and Lemeshow goodness of fit P=0.816) in the derivation cohort. External validation showed similar discrimination (C statistic 0.82 [95% CI, 0.72-0.91]) and calibration (Hosmer and Lemeshow goodness of fit P=0.866). CONCLUSIONS: Our internally and externally validated prediction model for sICH in patients with AIS who received intravenous thrombolysis may facilitate individualized prediction for intracerebral bleeding risk after intravenous thrombolysis for acute ischemic stroke.