Abstracts

Rationale: Analysis of self-reported data from online databases is a novel approach towards examining clinical aspects of seizure patterns and disease evolution. Clustered seizures are repetitive seizures separated by shorter than usual inter-seizure intervals(ISI) and pose special morbidity risks.We examined the clustering characteristics of self-reported seizures from patients with Tuberous Sclerosis Complex(TSC) entering data into SeizureTracker.com, a large free and publically accessible online seizure diary website.Methods: Data entered between 2007 - 2015 were reviewed. Subjects reporting TSC as “epilepsy etiology” were selected. The records include the subjects’ demographics and reported seizures.Subjects who reported only one seizure were excluded as this may represent an initial test entry. Three definitions of clustering were used: any three consecutive seizures in less than 24h, 12h and 6 hours respectively.The characteristics of clustering and the relationships between seizure duration, logarithmic transform of seizure duration, seizure type and triggers were assessed using dedicated statistics (T-test, Mann-Whitney test, Pearson’s Chi-square, regression analysis).Results: The 346 subjects reported 54194 seizures (2 to 3383 seizures, mean 156). Age ranged from 1 month to 52 years (285 children <18 years). Frequency of seizures type are indicated(Table1). For clustering analysis, outliers who reported greater than 1000 seizures were excluded, as were the seizures longer than 30 minutes(status epilepticus). The remaining 318 subjects reported 22927 seizures.The prevalences of clustering are indicated for each definition, also by age and seizure types(Table 2A,2B). The clustered seizures were shorter than the non-clustered seizures for all 3 definitions, p<0.001(Table2C). A trigger (change in medications, stress, fever, being sick, and/or tired /irregular sleep) was reported in association with 27.4% of the seizures.Two triggers were significantly associated with clustered seizures for all definitions: being sick and having irregular sleep (p<0.001) while reported stress or fever were associated with definition 2 and 3 (p<0.001)(Table2D).Conclusions: The prevalence of seizure clustering in subjects with TSC was high, ranging from 63% for clusters defined as 3 consecutive seizures in less than 24 hours to 45% for 3 seizures in less than 6 hours.Children were more likely to have clustered seizures than adults.Seizures most likely to cluster included atypical absences, infantile spasms, myoclonic and tonic seizures and focal seizures without dyscognitive features, while those less likely to cluster included focal seizures evolving bilaterally or with dyscognitive features and absence seizures. Clustered seizures were briefer than non-clustered seizures. Reported triggers associated with seizure clustering for varying definitions included intercurrent illness, having irregular sleep and stress. While most online diaries have known limitations including a lack of control entries, these results indicate that seizure clustering is a significant health concern among this population.