Abstracts

Clinical features of an epileptic seizure (seizure semiology) can provide important information regarding localization of the ictal onset zone. Understanding seizure semiology, and its concordance with EEG and imaging data, is particularly important in the evaluation of epilepsy surgery patients. The inter-rater reliability (IRR) of pediatric seizure semiology, with an emphasis on the order of appearance of clinical signs, has not been extensively studied. As part of a continuous quality improvement initiative in an epilepsy monitoring unit, we designed a form to record and code components of seizure semiology, and tested its usability and IRR., Based on expert opinion and a literature review of lateralizing signs in adults and children, a form was constructed listing lateralizing and nonlateralizing signs commonly seen in seizures. Four major categories [ndash] eye signs, face signs, trunk/limb signs, and other signs [ndash] were delineated. Within these 4 categories, specific elements were described and lateralized. Two investigators independently reviewed and coded the video records of 33 seizures from subjects aged 2 to 18 years. The primary outcome of interest was percent agreement when identifying the first three clinical signs, including lateralization when appropriate. Across all seizures, 95 clinical signs were assessed. Percent agreement corrected for chance agreement (Kappa statistic) was explored, but correction was minimal (less than 1%) due to the large number of possible choices., Agreement on behavioral state prior to the seizure occurred in 85% of records. Agreement on the presence of clinical signs occurred in 73/95 (77%) of clinical signs assessed. Lateralizing features were found in 65/95 (68%) of clinical signs assessed. In records with such features, agreement on lateralization occurred in 77% of records. Eighteen percent of lateralization disagreements (n=2) were due to rater errors. The other disagreements were disputes of whether signs were unilateral versus bilateral (n=7), or present or absent (n=2). All but two disagreements on lateralization were resolved in a consensus conference. We observed that 49% of initial signs had lateralizing features, while 76% of second signs were lateralizing, and 83% of third signs were lateralizing., We designed a practical, easy-to-use form for coding the semiological components of individual seizures recorded on video. IRR was good-to-excellent, and most lateralization disagreements were resolved on consensus conference. The second or third signs during a seizure were more often lateralizing than the first. Based on additional experience, this form will be optimized for greater ease-of-use and reliability. Future work could correlate the order of appearance of lateralizing signs with EEG data, imaging data and outcome in children being evaluated for epilepsy surgery.,