Prediction of epilepsy after stroke: Proposal of a modified SeLECT 2.0 score based on posttreatment stroke outcome

Objective: The SeLECT 2.0 score is a prognostic model of epilepsy after ischemic stroke. We explored whether replacing the severity of stroke at admission with the severity of stroke after treatment at 72 h from onset could improve the predictive accuracy of the score. Methods: We retrospectively identified consecutive adults with acute first-ever neuroimaging-confirmed ischemic stroke who were admitted to the Stroke Unit of the Ospedale Civile Baggiovara (Modena, Italy) and treated with intravenous thrombolysis and/or endovascular treatment. Study outcome was the occurrence of at least one unprovoked seizure presenting >7 days after stroke. Results: Participants included in the analysis numbered 1094. The median age of the subjects was 74 (interquartile range [IQR] = 64–81) years, and 595 (54.4%) were males. Sixty-five (5.9%) subjects developed unprovoked seizures a median of 10 (IQR = 6–27) months after stroke. The median values of the original and modified SeLECT2.0 scores were 3 (IQR = 2–4) and 2 (IQR = 1–3). The modified SeLECT 2.0 score showed better discrimination for the prediction of poststroke epilepsy at 36, 48, and 60 months after stroke compared to the original score according to the area under time-dependent receiver operating characteristic curves. The modified SeLECT 2.0 score had higher values of Harrell C and Somers D parameters and lower values of Akaike and Bayesian information criteria than the original score. The modified SeLECT 2.0 score produced more accurate risk predictions compared to the SeLECT 2.0 score at all evaluated time points from 12 to 60 months after stroke according to the Net Reclassification Index. Significance: Replacing baseline with posttreatment stroke severity may improve the ability of the SeLECT 2.0 score to predict poststroke epilepsy.