Identifying the pre-ictal state clinically would improve our understanding of seizure onset and suggest opportunities for new treatments. In our previous paper-diary study, increased stress and less sleep predicted seizures. Utilizing electronic diaries, we expanded this investigation. Variables were identified by their association with subsequent seizure using logit-normal random effects models fit by maximum likelihood. Nineteen subjects with localization-related epilepsy kept e-diaries for 12-14 weeks and reported 244 eligible seizures. In univariate models, several mood items and ten premonitory features were associated with increased odds of seizure over 12h. In multivariate models, a 10-point improvement in total mood decreased seizure risk by 25% (OR 0.75, CI 0.61-0.91, p=004) while each additional significant premonitory feature increased seizure risk by nearly 25% (OR 1.24, CI 1.13-1.35, p<001) over 12h. Pre-ictal changes in mood and premonitory features may predict seizure occurrence and suggest a role for behavioral intervention and pre-emptive therapy in epilepsy.
Copyright © 2012 Elsevier Inc. All rights reserved.