Abstracts

Rationale: The purpose of this study was to determine how the risk for seizures in acutely ill patients evolves as a function of abnormalities on continuous EEG (cEEG) monitoring. Previous work has shown that in patients without epileptiform abnormalities early in the recording, the risk of seizures in the subsequent 24 hours is low. However, one concern is that the delayed seizure rate in this population is still high. In the present study we quantify the temporal evolution of seizure risk in relation to cEEG findings over a long time period by generating models of the transitions between different cEEG states. These models are used to determine the expected incidence of seizures over time in patients with and without epileptiform abnormalities. Methods: We analyzed initial detection times of key EEG abnormalities (slowing, sporadic and periodic epileptiform discharges, and electrographic seizures) in 242 consecutive cases of acutely ill patients with at least 18 hours of continuous EEG (cEEG) monitoring. Baseline clinical and EEG variables were analyzed to identify predictors of subsequent seizures, and the transition rates over time for developing epileptiform abnormalities or seizures were estimated as a function of pathological features seen so far. These analyses were carried out using methods from survival analysis, including cumulative distribution functions, future event probability curves, and hazard functions. Results: Seizures occurred in 29% (n=70) of cEEG recordings. In 52 patients, the first seizure occurred within the first 30 minutes of monitoring. Of the remaining 190 patients, 63 had early epileptiform discharges, and 103 had no epileptiform abnormalities. Seizures were later detected in 22% (n=14) with early epileptiform discharges, vs 3% (n=3) without early epileptiform abnormalities (p < 0.001). Though seizure risk steadily declined throughout the course of monitoring, our projections suggest that even after a 24 hour seizure-free monitoring interval, patients with epileptiform abnormalities remain at moderate risk for seizures over the following 48 hours (nearly 4%); in contrast, patients without epileptiform discharges at 24 hours have a very low risk of developing seizures directly (<1%), though our analysis suggests a persistent moderate risk of developing delayed seizures indirectly via first transitioning to a higher risk group with epileptiform discharges (nearly 3%), then subsequently developing seizures. Conclusions: Seizure risk in acutely ill patients undergoing cEEG monitoring evolves substantially over time, and is strongly modulated by the nature of cEEG abnormalities detected early in the course of monitoring. This study presents a novel analysis of seizure temporal risk dynamics in acutely hospitalized patients undergoing cEEG monitoring. Substantial seizure risk stratification is possible on the basis of very early cEEG features. These findings provide guidance to allow more refined, patient-specific determination of the required duration of cEEG in critically ill neurological patients.