Abstracts

Rationale: Electrographic seizures (ESz) are a common in continuous EEG (CEEG) recordings in critically ill patients. This seizure activity can show a wide range of patterns, including differences in frequency, duration, and spatial extent. These various patterns may have different clinical implications for the patient, some requiring aggressive management while others may not. While factors such as etiology of seizures, co-morbid conditions, functional status and level of consciousness all play a role in guiding management of seizures in such patients, the actual pattern of seizure activity has implications for treatment as well. Here we report on our initial attempt to develop a seizure burden score (SBS) based on ESz parameters that could be used to summarize features of cEEG in a clinically meaningful way.Methods: Adult patients (age ≥ 18 years) who had at least one seizure during the first 4 hour of cEEG monitoring at Duke University Hospital from January 1, 2010 until August 22, 2012 were included in the study. 72 patients were included. Their EEG and patient charts were retrospectively reviewed for demographics/clinical paramters (age, co-morbid conditions, mental status at time of recording, length of ICU stay and hospitalization), estimates of clinical outcome (discharge disposition, and estimates of neurological function prior to hospitalization and at discharge, including Glasgow Coma Scale (GCS), Modified Rankin Scale (MRS) and Glasgow Outcome Scale (GOS)) and EEG characteristics during the first 4 hours of recording (best sustained background frequency, highest ictal discharge frequency, number of seizures/hour, average duration of individual seizures, total seizure time/hour, spatial extent of seizures). For analysis, outcome variables were classified into good/bad outcomes (for discharge location SNF, hospice or death considered bad; for others, a worsening change on the scales compared to baseline considered bad). EEG variables were classified as low/high using the median or 75%ile as the break point (for spatial extent, focal/hemispheric versus bilateral/generalized). Odds ratios (for a bad outcome for high value of EEG parameter compared to low) were calculated along with 95% confidence intervals and Pearson Chi-Square test was used to test for statistical significance.Results: Basic results are summarized in the Table (more detailed analysis will be presented). The EEG variables that correlated best with outcome were best background and highest ictal discharge frequency -- with higher frequencies associated with good outcomes. Other seizure related parameters showed only weak trends in association with outcome. Combining these parameters into an EEG seizure burden score did not increase the association.Conclusions: EEG paramaters that were most closely associated with outcome appear to be those that reflect the overall health of the brain (e.g., background frequency). EEG parameters traditionally associated with seizure severity (frequency, duration, spatial extent, and total seizure time) did not correlate with outcome in this retrospective study.