Abstracts

To determine whether a diagnosis of nonepileptic seizures (NES) can be predicted by sociodemographic information and a simple questionnaire. Epileptic and NES patients were recruited from a multispecialty community-based clinic equipped with an epilepsy-monitoring unit. The diagnosis of NES was confirmed by video-EEG monitoring and clinical data. Patients completed a self-administered survey consisting of quality of life (QOLIE-31), depression, and anxiety scales. Subsequently, an interviewer-based questionnaire was used to assess sociodemographic information, attitudes toward self-care, social-support, trust in the healthcare system, self-efficacy, and health-literacy. Epilepsy-specific parameters included seizure frequency (per month), seizure control (seizure-freedom for the past 12 months), time since diagnosis, and AED compliance. The bivariate relationship between each potential predictor and the diagnosis of NES was tested using the chi-square test for categorical variables and the independent [italic]t[/italic]-test for continuous variables. Predictors with [italic]p[lt][/italic]0.20 in the bivariate analysis were entered into a stepwise conditional multivariate logistic regression model with the diagnosis of NES as the dependent variable. Predictors with [italic]p[/italic][gt]0.05 in the regression model were sequentially deleted from the analysis until only statistically significant predictors remained. To date, 98 epileptics (mean age 42+/-17, 60.2% female, mean seizure frequency 2.6+/-0.7) and 19 NES (mean age 41+/-14, 94.7% female, mean seizure frequency 11.1+/-3.7) have completed both surveys. Bivariate analysis identified female gender ([italic]p[/italic]=0.004), Non-Caucasian ethnicity ([italic]p[/italic]=0.087), unemployment ([italic]p[/italic]=0.102), low income ([italic]p[/italic]=0.016), increased seizure frequency ([italic]p[/italic][lt]0.001), greater time since diagnosis ([italic]p[/italic]=0.017), poor AED compliance ([italic]p[/italic]=0.196), low QOLIE-31 score ([italic]p[/italic]=0.001), depression ([italic]p[/italic]=0.001), anxiety ([italic]p[/italic]=0.001), low social support ([italic]p[/italic]=0.067), and low self-efficacy ([italic]p[/italic]=0.068) as potential predictors of a diagnosis of NES. The [italic]medication effects[/italic] subscale of the QOLIE-31 and the [italic]tangible[/italic] subscale of the social support inventory were the only subscales not found to be significantly related to a diagnosis of NES. The final logistic regression model retained female gender (OR=17.3 [1.6, 183.6]), low income (graded from 1=lowest to 7=highest, OR=0.74 [0.56, 0.99]), low [italic]overall-quality-of-life[/italic] subscale score of QOLIE-31 (scaled 0-100, OR=0.94 [0.92, 0.98]), and higher seizure frequency (OR=1.1 [1.03, 1.17]) as independent predictors of a diagnosis of NES. The C-statistic and GOF statistic for the model were 0.894 and 0.338, respectively. The positive predictive value for predicting a diagnosis of NES in our patients was 91.4%. 2 items from the QOLIE-31 questionnaire ([italic]overall-quality-of-life[/italic] subscale) and sociodemographic information may be useful in differentiating NES patients from true epileptics.