By Kimberly Pargeon, MD

Assistant Professor of Clinical Neurology, Weill Cornell Medical College

Dr. Pargeon reports no financial relationships relevant to this field of study.

SYNOPSIS: Investigators prospectively analyzed 72-hour continuous electroencephalograms to identify clinical and electroencephalogram risk factors for having seizures and developed a model for “time-dependent” seizure risk. Electrographic seizures occurred in 23% of all patients. The only significant clinical predictors of seizures were presence of coma and prior clinical seizure history.

SOURCE: Struck AF, et al. Time-dependent risk of seizures in critically ill patients on continuous electroencephalogram. Ann Neurol 2017; July 6. [Epub ahead of print].

Non-convulsive seizures (NCS) are reported in about 20% of critically ill patients,1 typically with little to no overt clinical signs, thus requiring identification with continuous electroencephalogram (CEEG).2 With recent studies suggesting that NCS can have significant adverse effects on the brain, some current practice guidelines recommend ordering CEEG in critically ill patients who are at increased risk for seizures and with unexplained altered or fluctuating mental status.2 However, the question always remains about how long monitoring should be continued, particularly in locations where resources are limited. A commonly cited study from 2004 showed that nearly 90% of NCS in critically ill patients were recorded within the first 24 hours,1 but this study did not discern specifically which patients were at greatest risk.

Struck et al developed a model for “time-dependent” electrographic seizure risk in critically ill patients based on clinical risk factors and CEEG abnormalities. At two academic hospitals, 665 eligible critically ill patients were monitored prospectively on CEEG for 72 hours. One of the three authors read and scored the CEEGs after undergoing a certification test. Both the presence and time to emergence of seizures, as well as rhythmic or periodic patterns, were recorded. Clinical characteristics, such as gender, presence of brain injury, history of epilepsy or recent acute seizures, coma, and presence of focal neurological findings, were recorded prospectively.

Electrographic seizures occurred in 151 of 665 (23%) patients. The only clinical risk factors that were significant independent predictors for seizures were 1) prior clinical seizures (history of epilepsy and/or presence of acute clinical seizures), and 2) coma. In terms of EEG findings, lateralized periodic discharges (LPDs), lateralized rhythmic delta activity (LRDA), and brief potentially ictal rhythmic discharges (BIRDs) were significantly associated with electrographic seizures, whereas sporadic epileptiform discharges, bilateral independent periodic discharges, and lateralized rhythmic spike-and-wave were not statistically significant. The median observed time of EEG “risk pattern” emergence (i.e., any epileptiform abnormality) was four minutes and of seizure emergence was 44 minutes. However, using a “multistate survival analysis,” the authors attempted to identify the “decaying” risk of seizures as a function of EEG duration, abnormal EEG findings, and clinical risk factors. They noted that seizure risk declined quickly if EEG abnormalities were not seen and there were no clinical risk factors. As an example, if a patient had no seizure history and was not comatose, a negative routine EEG of at least 30 minutes would be sufficient to place seizure risk at < 5% over the next 72 hours. However, if there was an EEG abnormality or if there was one or more clinical risk factor, patients would need at least 15-44 hours of CEEG.


Critically ill patients can have multiple reasons for altered or fluctuating mental status, but given that nearly 20% can have NCS, CEEGs are ordered with increasing frequency in ICUs. However, some facilities have limited resources, so having guidelines would be meaningful. In this study, 23% of patients had electrographic seizures, comparable to previous studies. What is most noteworthy about this study, however, is looking at seizure risk in a “time-dependent” manner. If a patient had no history of previous clinical seizures and was not comatose, this study suggested that routine EEG may be sufficient to rule out seizures for unexplained altered or fluctuating mental status, assuming no epileptiform abnormalities are seen. However, if any of these EEG abnormalities are appreciated, particularly LPDs, LRDA, or BIRDs, within the first 30 minutes of recording or if a patient has one of these two clinical risk factors, then EEG should be extended to at least 24 hours. If a patient has both EEG “risk patterns” and one of these clinical risk factors, then the authors recommended extending the recording to at least 48 hours.


  1. Claassen J, et al. Detection of electrographic seizures with continuous EEG monitoring in critically ill patients. Neurology 2004;62:1743-1748.
  2. Struck AF, et al. Time-dependent risk of seizures in critically ill patients on continuous electroencephalogram. Ann Neurol 2017; July 6. [Epub ahead of print].