Abstracts

To differentiate epileptic (ES) and psychogenic non-epileptic seizures (NES), psychological testing is often performed as an adjunct to long-term video-EEG monitoring (VEEG). The Minnesota Multiphasic Personality Inventory (MMPI) is the most commonly used measure for this purpose. Correct overall classification rates of 70% may be expected using the decision rules of Wilkus et al. (1984). In this study, we examined whether combining MMPI profiles with other medical and psychiatric history variables may enhance diagnostic accuracy as compared to classification based on personality profiles alone. Subjects were drawn from a consecutive series of adult patients who underwent VEEG to characterize seizures. Those with a definite diagnosis of NES or ES based on ictal EEG findings and clinical semiology were included. Excluded patients had documented concurrent NES and ES, only subjective spells, or did not complete the MMPI-2. 105 patients with NES and 109 patients with ES met these criteria. The MMPI classification rules of Wilkus et al. were applied without modification to MMPI-2 profiles. Routine medical and psychosocial histories were abstracted on standardized data sheets. Multiple variables were recorded, including age, sex, age at onset and duration of seizure disorder, number of other current and past medical diagnoses, number of current medications, presence of chronic pain disorder, number of psychiatric conditions, and history of physical or sexual abuse. Stepwise logistic regression was used to identify the variables giving the best diagnostic classification. Significant differences were observed between the ES and NES groups on all of the aforementioned variables. MMPI profiles correctly classified 87% of the NES subjects, and incorrectly classified 28% of the ES patients, for an overall correct classification rate of 79%. A total correct classification rate of 82% was acheived by a combination of 6 variables that were more closely associated with NES: Female sex, short duration of seizure disorder, history of chronic pain, higher number of past and current medical diagnoses, and the total number of documented psychiatric problems. The highest overall accuracy was 85.5% (86% sensitivity for NES, 85% specificity) when these clinical variables were combined with the MMPI data. Differentiating NES from ES patients on the basis of psychometric measures such as the MMPI may be useful as an adjunct to other diagnostic studies, but the probability of making false positive errors (classifying ES patient as NES) based on this method alone is a significant limitation. Classification accuracy is enhanced when other simple clinical data are considered along with MMPI scores.