Abstracts

Rationale: Reliable determination of prognosis in post-cardiac arrest adults treated with therapeutic hypothermia (TH) is difficult for a number of reasons including: TH itself, sedatives and neuromuscular blockers used in the majority of TH protocols, and the inability to rely on pre-hypothermia era outcome studies. Continuous EEG monitoring (cEEG) may identify reliable patterns present in the first few days post-arrest that predict neurologic outcome. Hypothesis: Background activity and a subset of epileptiform activity can be abstracted from daily cEEG reports to build a prediction algorithm for short-term neurologic outcome.Methods: We retrospectively collected daily cEEG interpretations and clinical data including best neurologic outcome (Cerebral Performance Category [CPC]) prior to hospital discharge from medical records of consecutive comatose adult patients treated with TH post-cardiac arrest at 5 academic medical centers (2006-2010). A tiered EEG score was created: (0) best - alpha to theta; (1) indeterminate - theta to delta & alpha coma; and (2) worst - no clear cerebral activity, burst-suppression, or GPEDs on severe voltage attenuated background. Presence of frequently appearing periodic or rhythmic sharp/spike activity (PD, interictal or ictal) was incorporated into each of these tiers.Results: 324 consecutive patients were included as having some form of EEG during the first 3d post-arrest.. Pertinent demographics and monitoring data: mean age 60 y (sd 17, range 19-96 y); 99% had cEEG; median days of cEEG/patient was 3 (range 1-21); Outcome: 27% had good neurologic outcome (CPC 1&2), 11% had recovery of awareness with poor neurologic function (CPC3), and 62% were dead or vegetative (CPC4&5).PDs were present in 116 (36%) of the patients, 62 of whom also had electrographic seizures An additional 8 patients had electrographic seizures and no PDs; i.e. 22% of the total cohort was noted to have seizures. The 3-tiered EEG background frequency and adjunct PDs scale was highly associated with poor neurologic outcome for each day of EEG (EEG score OR 8.5 & PD OR 18.5; p<0.00001 logistic regression chi-squared statistic). cEEG during 24-48h & 48-72h (Table 1) best predicted neurologic outcome. Worst EEG score (#2) performed well in predicting poor outcome, and addition of PD improved prediction of poor outcome in some cases. EEG score 2 (48-72h) predicted poor outcome with a FPR of 0%[95% C.I. 0-5.5%].Conclusions: This large multicenter retrospective study found that in patients post-arrest treated with TH, cEEG in the first few days can predict good and poor neurologic outcome via a simple 3-tiered rating scale combined with the presence or absence of periodic/rhythmic sharp/spike activity. A poor cEEG score in the first 72h is a strong predictor of poor outcome. This should be confirmed in an independent cohort and combined with other clinical data to maximize accuracy of post-arrest prognostication.