Predicting psychogenic nonepileptic seizures with the Personality Assessment Inventory and seizure variables

Early identification of likely cases of psychogenic nonepileptic seizures may prompt earlier and more appropriate diagnosis and treatment. This study investigated the validity of multiple independent self-report variables in predicting psychogenic nonepileptic versus epileptic seizures. One hundred forty-three patients with psychogenic nonepileptic seizures and 129 patients with epileptic seizures were compared with the Personality Assessment Inventory, age at seizure onset, years since first seizure, and length of seizures. Diagnostic prediction was made by analyses of variance and logistic regressions. Cut scores were developed for the multivariate predictors with maximal sensitivity and specificity. The Personality Assessment Inventory conversion subscale (SOM-C), years since first seizure, and length of seizures provided 84% correct classification. Ninety-six percent of patients meeting cut scores of SOM-C ≥70, years since first seizure ≤8, and length of seizures ≥3 minutes were diagnosed with psychogenic nonepileptic seizures. Results demonstrated good ability to identify likely cases of psychogenic nonepileptic seizurs by multiple predictors that are easily and cost-effectively obtainable.