Development and validation of a predictive outcome score of cerebral venous thrombosis.

Background and Purpose: Cerebral venous thrombosis (CVT) is a rare disease with a variable outcome. The aim of this study was to develop and validate a predictive outcome score for CVT patients.
Methods: The score was based on the 8 predictive variables of poor outcome (modified Rankin Scale score>2) from the International Study on Cerebral Vein and Dural Sinus Thrombosis: age >37, male, mental status disorder, coma, intracranial hemorrhage, deep CVT, CNS infection, and malignancy. To assign a weighted index (WI), the natural logarithm of the hazard ratio of each variable was calculated, multiplied by 10 and rounded to the nearest integer. The individual score of each patient was the sum of the WI. The validation sample consisted of a retrospective single center cohort of 90 CVT patients.
Results: Sixteen patients (18%) had a poor outcome. The predictive score had an area under the receiver operating characteristic curve of 0.81 (95% CI 0.71-0.90). The cut-off score with the maximum sum of sensitivity and specificity was a score> or =14 with sensitivity of 88% (81%-95%) and specificity of 70% (61%-79%). The predictive value of a score<14 for good outcome was 96% (92%-100%), whereas the predictive value of a score> or =14 for poor outcome was 39% (29%-49%).
Conclusions: This relatively simple predictive outcome score may be useful in CVT patients. A cut-off score of 14 reliably predicts good outcome, but is less accurate in predicting poor outcome.