LAUSR

Approximately one-third of epilepsy patients are medically refractory and could benefit from surgery. Currently, it is difficult to predict precisely the postoperative likelihood of seizure freedom. Here, we sought to develop a regression-based clinical score to predict the likelihood of postoperative seizure freedom separately for pediatric and adult patients. All patients undergoing surgery at the adult (July 2004-January 2017) and pediatric institutions (July 2003-July 2014) were reviewed. Demographics, seizure-specific history, EEG, imaging, and pathology data were collected. The primary outcome was 5-yr seizure freedom (Engel). Random selection of 70% of the cohort was used for development while the remaining 30% was used for validation. The proportion for the odds ratio obtained from the regression model was used to construct whole number coefficients for each significant variable. Receiver operating characteristic (ROC) curves were used to evaluate model utility. A total of 194 adult and 316 pediatric subjects (Total N = 510) were included. For the pooled cohort, the score range was 0 to 18 (absence of GTC = 4, monthly seizures = 20 = 4, MRI abnormality = 5, Temporal lobe epilepsy = 3, not needing iEEG = 2). For adults, the score range was 0 to 9 (constant = 2, absence of GTC = 3, no epilepsy family history = −2, MRI abnormality = 2, not needing iEEG = 2). For pediatric patients, the score range was 0 to 8 (absence of GTC = 3, monthly seizures = 20 = 3, not needing iEEG = 2). The AUC of the predicted 5-yr seizure-freedom for each score compared to the actual rate was >0.79 in all cases. Clinical prediction tools, developed separately for pediatric and adult patients, can be used for patient selection and counseling. Further multicenter studies are needed to validate this score while comparing its performance to other available scores and nomograms.