Seizure prediction and recall.

Using separate generalized mixed-effects models, we assessed seizure recall and prediction, as well as contributing diagnostic variables, in 83 adult patients with epilepsy undergoing video/EEG monitoring. The model revealed that when participants predicted a seizure, probability equaled 0.320 (95% CI: 0.149-0.558), a significant (P<0.05) increase over negative predictions (0.151, 95% CI: 0.71-0.228]). With no seizure, the rate of remembering was approximately 0.130 (95% CI: 0.73-0.219), increasing significantly to 0.628 (95% CI: 0.439 to 0.784) when a seizure occurred (P<0.001). Of the variables analyzed, only inpatient seizure rate influenced predictability (P<0.001) or recollection (P<0.001). These models reveal that patients were highly aware of their seizures, and in many cases, were able to make accurate predictions, for which seizure rate may be an important factor. Copyright (c) 2010 Elsevier Inc. All rights reserved.