Abstracts

Rationale: A number of studies have reported associations between premonitory symptoms and short term seizure incidence. However, most such studies have relied on retrospective recall of premonitory symptoms rather than systematic prospective data collection, raising concern regarding possible bias. Methods: 19 patients with a history of 3+ seizures per month completed twice-daily electronic diaries for up to five months. Participants were asked about a number of premonitory symptoms though questions of the form, “Are you experiencing any of the following” followed by multiple choice menus, with an opportunity to add up to three open-ended responses following a multiple choice response of “Other”. Associations between the morning diary entries and the occurrence of a seizure within 12 hours of the diary entry was examined using logit-normal random effects models, taking into account the fact that each individual contributed multiple observations to the dataset. Results: Participants reported a total of 268 seizures. Symptoms were reported on 1680 morning diaries (range per patient 46 to 150). We specifically collected data on 18 premonitory features, and of those, 10 were significantly associated with seizure occurrence (Table 1.) Blurred vision (OR 18.20, 95% CI 7.37, 44.97), Light Sensitive (OR 8.35, 95% CI 2.46, 28.35), and Emotional (OR 5.67, 95% CI 1.96, 16.42) showed the strongest association. Conclusions: Using time stamped data from electronic diaries, premonitory features are significant predictors of seizure occurrence in univariate analysis. Future work will focus on developing individual and population level multivariate models for seizure prediction within a clinically meaningful time window.