Author ORCID Identifier

Date Available


Year of Publication


Degree Name

Doctor of Philosophy (PhD)

Document Type

Doctoral Dissertation


Arts and Sciences



First Advisor

Dr. Frederick A. Schmitt


The published literature on the Personality Assessment Inventory (PAI) for psychogenic nonepileptic seizure (PNES) diagnosis includes a variety of interpretation methods to distinguish PNES from epileptic seizures (ES) and offers mixed findings. The purpose of this study was to use a cross-validation approach to create and derive new decision rules for the PAI to best differentiate PNES from ES. Data from 773 patients (PNES n = 328, ES n = 445) who underwent long-term video EEG (vEEG) monitoring and completed a PAI were examined. Individuals with invalid PAI profiles were removed, and patients were randomly assigned to the “development” group (DEV) or the “application” group (APP). Receiver operating characteristic (ROC) curves with DEV demonstrated the best cut score for each scale of interest. ROC curves were repeated with APP. Additional analyses examined the utility of sequential decision rules incorporating multiple scales. Of the individual scales, SOM-C demonstrated the best diagnostic accuracy (sensitivity [SN] = 60.7%, specificity [SP] = 81.3%) at a cut score of T ≥ 75. Cross-validation with APP confirmed this cut score outperformed other cut scores (positive predictive value [PPV] = 67.2%, negative predictive value [NPV] = 76.1%), as well as other decision rules presented in the literature. Additional analyses examining sequential decision rules with SOM-C ≥ 75 or SOM-C = 70-74 with SOM-S ≥ 65 demonstrated the highest predictive power (PPV = 73.2%, NPV = 79.1%). The results of this study demonstrate a new and effective method for using the PAI as a screener to distinguish PNES from ES. Utilization of these decision rules can assist clinicians in determining appropriateness of and immediate need for vEEG monitoring for diagnostic clarification.

Digital Object Identifier (DOI)