BackgroundThere is a need for the development of brief tools to screen for cognitive impairments in epilepsy patients in order to prioritize and direct formal comprehensive cognitive testing. Yet, shorter cognitive screening tools are limited in their breadth of cognitive domains or have not been intensively studied on an epilepsy population. This study used a brief cognitive screening tool in order to compare cognitive profiles between patients with epilepsy and those with nonepileptic seizures. MethodsPatients admitted to the Royal Melbourne Hospital video-EEG monitoring unit between 2005 and 2017 were included. Patients were categorized according to seizure etiology (epileptic, psychogenic or other nonepileptic seizures), epilepsy syndrome (focal or generalized; temporal lobe (TLE) or extra-temporal lobe epilepsy (ETLE)), seizure frequency, and anti-seizure medications (ASMs). Attention, visuoconstructional, memory, executive, and language functioning were assessed with the Neuropsychiatry Unit Cognitive Assessment Tool (NUCOG). General linear mixed models were computed to investigate cognitive profiles according to diagnostic group and other clinicodemographic variables. Results800 patients were included in the analysis (61% female and 39 % male, median age 36 years). Patients with both epileptic seizures and psychogenic seizures (n = 25) had the lowest total scores on NUCOG, followed by patients with epileptic seizures (n = 411), psychogenic seizures (n = 185), and nonepileptic seizures (n = 179, p = 0.002). Specifically, patients with epileptic seizures performed worse than those with nonepileptic seizures in the executive, language, and memory domain, and had lower language domain scores than those with psychogenic seizures. Patients with bilateral TLE had poorer performance than those with unilateral TLE, particularly for memory function. Specific ASMs and polypharmacy but not seizure frequency had a negative effect on cognition (p < 0.001). NUCOG scores did not differ between focal and generalized epilepsies, or between TLE and ETLE. ConclusionThe NUCOG differentiated cognitive profiles in patients with uncontrolled seizures due to different etiologies. Bilateral TLE and medication adversely affected cognitive performance, and overall patients with epilepsy performed worse than those with nonepileptic seizures. These results provide further evidence for sensitivity of the NUCOG for detecting cognitive impairment in patients with seizure disorders.