Abstracts

Rationale: Prolonged EEG monitoring in critically ill patients has revealed a high incidence of seizures. This can range from continuous, generalized seizure activity to intermittent seizures of variable duration, frequency, and spatial extent. The significance of these patterns with respect to need for treatment, potential for additional neurological injury and outcome is uncertain. The purpose of this study is to assess the impact of different seizure characteristics as a tool for describing severity and intensity of seizures as defined by outcome.Methods: Prolonged EEGs performed at Duke University Hospital from 1/1/10 to 8/22/12 were reviewed to find patients who had seizures within the first 4 hours of recording. Exclusion criteria were age <1 year, unavailability of raw EEG, and prolonged EEGs performed for elective characterization of spells. Data collected included demographic parameters and relevant history, outcome at discharge, mental status during EEG, lengths of ICU and hospital stay. Seizure characteristics were determined by reviewing the raw EEG. This included onset and offset of individual seizures, number of seizures, number of electrodes involved, ictal discharge frequency, and interictal background frequency. Independent variables included seizure duration, seizures/hour, spatial extent of seizures, clinical activity during seizures, and [SBSTE = seizure time/hour*spatial extent] and [SBSTED = seizure time/hour*spatial extent*average seizure duration]. Patients were then categorized according to whether or not the variables was above or below the median and held dependent to outcome.Results: 97 patients were identified. Age ranged from 1-95 years old, 49 were female. Forty patients had epilepsy, 23 brain tumor, 23 stroke, 14 intracranial hemorrhage, and 11 hypoxic insult prior to EEG monitoring. For discharge outcome, 49 home, 8 rehabilitation, 14 long term care, and 26 hospice care or died. Total seizure time ranged from 17 sec to over 3 h, with median of 12 min. The total number of seizures/patient ranged from 1 to 291 with a median of 1.8 seizures/hour. Average seizure duration for individual patients ranged from 7 sec to almost 2 h, with a median seizure duration among all patients of 60 sec. With regards to the equations, SBSTE demonstrated a logarithmic median at 2.51 (range: 0.88-3.74) and SBSTED demonstrated a logarithmic median at 4.3 (range: 2.0 to 7.31). None of the seizure characteristic variables predicted outcome with any statistical significance.Conclusions: No statistical significance was seen in this preliminary study. The data indicate that the current combination parameters tested do not seem to have be predictive of overall patient outcome. Further testing looking at other variables associated with seizures and seizure complications and possibly focusing only on ICU patients may provide us with better parameters from which to predict prognosis in patients with seizures. Another option may be to use a different measure of seizure severity instead of overall outcome, which is clearly impacted by many other clinical variables.