Abstracts

Rationale: Video-EEG monitoring (VEEG) helps detect seizures in the intensive care unit (ICU) setting where subclinical seizures are thought to be common. Yield of VEEG in this setting is not completely defined and factors associated with detection of seizures are not clear. Recent widespread use of therapeutic hypothermia and anesthetic drips may influence yield.Methods: Reports of VEEG studies performed in Medical, Cardiac and Neurological ICUs at a large referral hospital between January 2010 and October 2012 were reviewed by board certified electroencephalographers. We only included studies prompted by concern about undiagnosed or persisting seizures. Demographic and clinical information as well as the occurrence of seizures, interictal epileptiform discharges (IEDs), periodic lateralizing epileptiform discharges (PLEDs), and generalized periodic epileptiform discharges (GPEDs) were databased. Definitions of these items followed Neurology 2004;62:1743-1748. All original studies were screened by board certified EEG technologists with additional training in ICU VEEG as well as commercially available seizure detection software. All studies were further reviewed by board certified electroencephalographers.Results: 490 individuals (248 women, average age 58.2 years, range 17-96) underwent 503 ICU VEEG studies totaling 1828 days of VEEG (average 3.6 days/study, range 1-30). One or more seizures were recorded during 89/503 (17.9%) of VEEG studies. Seizures were recorded during 243/1828 (13.3%) days of monitoring. In studies with seizures, first seizure was detected by the end of day 1 of VEEG in 59/89 (66%), the end of day 2 in 82/89 (92%), and the end of day 4 in 89/89. Subclinical seizures alone were recorded in 44/89 (49%) of studies with seizures and 44/503 (8.7%) of all studies. At least some subclinical seizures were recorded in 65/89 (73%) of studies with seizures and 65/503 (12.9%) of all studies. IEDs were recorded in 112/503 (22.3%), GPEDs in 46/503 (9.1%), and PLEDs in 39/503 (7.8%) of the studies. Suspicious paroxysmal clinical events that were not epileptic were marked by caregivers in 137/503 (27.2%) of studies. Seizures were more likely during studies when studies were immediately preceded by clinical seizures or when IEDs, PLEDs, or GPEDs were recorded during the study (univariate analysis, Table 1). Binary logistic regression adjusting for duration of VEEG found that seizure occurrence during a study was independently associated with the recording of IEDs or GPEDs during that study (Table 2). Therapeutic hypothermia was used in 52 studies. Number of studies with seizures, and number of seizures per study did not differ from those without hypothermia.Conclusions: Seizures commonly occur during ICU VEEG. Approximately half are subclinical. Our experience suggests that yield of recording first seizure drops significantly following two days of ICU VEEG. Non-epileptic, seizure-like events often occur in the ICU setting; VEEG distinguishes between these and epilepsy. IEDs and GPEDs are commonly recorded and both are associated with seizures in the ICU setting.