Yield of repeated intermittent EEG for seizure detection in critically ill adults

Abstract

Seizures are common in critically ill patients and prevalence can exceed 30% in the neuro-intensive care unit (ICU). Continuous EEG monitoring (cEEG) is the gold standard for seizure detection in critically ill patients. To determine the yield of intermittent EEG (iEEG) to detect critically ill adult patients with seizures and to identify the factors that affect this yield. We retrospectively analyzed cEEG data and medical records from 977 consecutive critically ill patients undergoing cEEG. We included those presenting at least one electrographic seizure during the first 24hours of cEEG. Patients with hypoxic-ischemic encephalopathy were excluded. For seizure detection, we reviewed six 30-minute epochs on cEEG selected at H0, H3, H6, H12, H18 and H24. Seizures occurred in 10.75% (105/977) of patients. Level of consciousness was impaired in 79 (75%) of patients, with 42 (40%) in coma. Review of the H0 epoch on cEEG permitted to detect seizures in 61 (58%) patients. These figures increased to 70 (67%), 75 (71%), 91 (87%) and 97 (92%) patients for a sampling every 24, 12, 6 and 3hours, respectively (P=0.02). Frequency of seizures on cEEG was the only factor significantly affecting the probability of seizure detection. Sampling every 6hours revealed seizures in all patients with more than six seizures per 24hours. iEEG repeated every 6hours can accurately detect patients presenting seizures, especially when seizure frequency is greater than six per 24hours. These findings have practical implications for electrographic seizure detection in critically ill patients in settings lacking cEEG.