The semantics of epileptic and psychogenic nonepileptic seizures and their differential diagnosis

The aim of the study was to highlight the fundamental linguistic elements through a simplified scoringtable to build a simplified and easier diagnostic model for linguistic evaluation by nonexperts.Methods:The study was based on interviews performed in patients recruited at the University of Milano Bicocca,at the Epilepsy Center of the University of Messina, and at the Epilepsy Monitoring Unit of the University of Roch-ester, who experienced“seizures”and underwent video-electroencephalogram (vEEG) for differential diagnosis.All enrolled subjects underwent a video-recorded interview consisting of afixed sequence offive questions. Sub-sequently, a researcher examined the video recordings, blind to the vEEG results, andfilled in the simplified lin-guistic evaluation (SLE) scoring table. The best cutoff score for the diagnosis was determined using a ROC curveand was selected as the value leading to the maximum number of patients correctly classified.Results:The study sample consisted of 35 interviews.The receiver operating characteristics (ROC) curve analysis showed that the best cutoff for the diagnosis was 12.The accuracy was 82.9% for this cutoff value. The area under the ROC curve (AUC) was 0.81 (95% confidence in-terval (CI): 0.66–0.97). Classifying patients using the SLE scoring table as diagnostic tool, with the selected cutoffscore, 17 out of 19 patients with psychogenic nonepileptic seizures (PNES) were correctly classified (89.5% sen-sitivity), while 12 out of 16 patients with epileptic seizures (ES) were correctly classified (75% specificity). Posi-tive and negative predictive values are, respectively, 81% and 85.7%.Conclusions:Despite some limitations, the use of the SLE scoring table may reduce costs, and conversation anal-ysis (CA) might help achieving a timely and reliable diagnosis of PNES