Abstracts

Rationale: In major medical centers, patients in prolonged (>7 days) refractory status epilepticus (PRSE) are encountered several times per year. Beyond the first standard treatments (benzodiazepines, phenytoin, barbiturates) the choices of medications are dictated by personal experiences and preferences rather than data. The predictive values of EEG and neuroimaging features for outcome are not clear. This multicenter retrospective study aimed to evaluate these clinical features in relatively large sample of patients in PRSE (data on treatment and outcomes are presented separately). Methods: Three academic centers within the Critical Care EEG Monitoring Research Consortium systematically identified and reviewed patients in PRSE over the period of 10 years (2000-2010, inclusive). Clinical, laboratory, and outcomes (MRS Modified Rankin Score) data were reviewed using standardized forms and definitions. PRSE was defined as continuous or intermittent seizure activity without return to baseline between seizures despite treatment with anesthetic infusion for at least seven days. IRB approval was obtained prior to data collection at each of the centers. For the purpose of this study, only EEG and neuroimaging data are discussed. Data on 63 PRSE patients were included. Patients with PSE resulting from cardiopulmonary arrest were excluded for the purposes of this study. Data were analyzed using Fisher exact tests and logistic regression.Results: EEG data: the presence (or absence) of interictal epileptiform discharges or periodic discharges was not significantly different in those with good outcomes (MRS 0-4) vs. poor outcomes (MRS 5-6; all p>0.05). Initial EEG reactivity showed a trend towards significance with better outcomes in patients with reactive EEGs (p=0.0664 for MRS 0-3 and p=0.099 for MRS 0-4; both vs. MRS 5-6). Normal initial CT predicted good outcome (p=0.0401 for MRS 0-2; p=0.0071 for MRS 0-3; p=0.0874 for MRS 0-4; all vs. MRS 5-6). MRI results had overall no predictive value but stable MRI results (no evolution) showed a trend towards better outcomes (p=0.071 for MRS 0-4 vs. MRS 5-6).Conclusions: Overall, the results of this retrospective study show that certain aspects of EEG and neuroimaging may be useful tools in predicting short- and long-term outcomes in patients in PRSE. Initial EEG reactivity and initially normal head CT are associated with better outcomes.