Abstracts

Rationale: Misdiagnosis of epilepsy is common and is difficult to undo. Non-epileptic seizures (NES) and epilepsy are treated differently and more than 75% of patients with NES are inappropriately treated with anticonvulsant medications. Treating NES as epilepsy can cause significant harm to the patient and outcome is better if it is recognized early. The mean time from onset of NES to diagnosis is about 7 years. Video EEG (VEEG) is the currently accepted gold standard, is a highly specific test for epilepsy, and can definitively differentiate between NES and epilepsy. However, VEEG is a poor screening test. A good screening test is quick, safe, inexpensive, easy to use, widely available, acceptable to the patient, and has high sensitivity. Many potential tests have been reported in the literature but none meet enough of these criteria. The objective of this study was to generate a formula based on common observations of patients with NES, and evaluate this formula as a screening test for differentiating patients with NES from those with epilepsy.Methods: All adult patients completing scalp VEEG monitoring at the UAB Epilepsy Center during a six month period and diagnosed with epilepsy (78), NES (78), or both (7), were included. Minors, prisoners, and mentally retarded patients were excluded. A retrospective chart review recorded each subject’s final diagnosis, age, gender, age at onset, results of MRI and routine EEG, seizure frequency, number of: drug allergies, CNS medications, AEDs tried in the past, psychogenic diagnoses. Best-subsets multiple logistic regression was performed to generate a formula based on the variables most predictive of the diagnosis. The results were cross-validated and sensitivity and specificity were calculated. VEEG was used as the gold standard.Results: The variables found to be predictive of the diagnosis were age, age at onset, reported (unverified) EEG results, and the total number of drug allergies and psychogenic diagnoses. Variables not predictive were the gender, seizure frequency, MRI result, and the number of CNS meds and AEDs tried. When the formula was generated from patients monitored during the first 3 months, and then validated with a different group of patients seen the following 3 months, the sensitivity for diagnosing epilepsy was 100% and the specificity for epilepsy was 83%. When this order was reversed, the sensitivity was 97% and specificity was 92%. When the formula was generated from all 163 patients, and then validated with the same group of patients, the sensitivity was 98% and the specificity was 84%.Conclusions: With the high sensitivity reported above, this formula can meet all the criteria for a good screening test. If this test is further validated, patients identified as having NES can be referred for psychological evaluation - before incurring the cost in time and money, the inconvenience of VEEG monitoring, and the exposure to unnecessary drugs - and can still be monitored at a later time. This formula can serve as an additional tool for general practitioners and neurologists in evaluating this challenging patient population.