Residual Seizure Rate of Intermittent Inpatient EEG Compared to a Continuous EEG Model

Abstract

Background: Subclinical seizures are common in hospitalized patients and require electroencephalography (EEG) for detection and intervention. At our institution, continuous EEG (cEEG) is not available, but intermittent EEGs are subject to constant live interpretation. As part of quality improvement (QI), we sought to estimate the residual missed seizure rate at a typical quaternary Canadian health care center without cEEG. Methods: We calculated residual risk percentages using the clinically validated 2HELPS2B score to risk-stratify EEGs before deriving a risk percentage using a MATLAB calculator which modeled the risk decay curve for each recording. We generated a range of estimated residual seizure rates depending on whether a pre-cEEG screening EEG was simulated, EEGs showing seizures were included, or repeat EEGs on the same patient were excluded. Results: Over a 4-month QI period, 499 inpatient EEGs were scored as low (n = 125), medium (n = 123), and high (n = 251) seizure risk according to 2HELPS2B criteria. Median recording duration was 1:00:06 (interquartile range, IQR 30:40–2:21:10). The model with highest residual seizure rate included recordings with confirmed electrographic seizures (median 20.83%, IQR 20.6–26.6%), while the model with lowest residual seizure rate was in seizure-free recordings (median 10.59%, IQR 4%–20.6%). These rates were significantly higher than the benchmark 5% miss-rate threshold set by 2HELPS2B (p<0.0001). Conclusions: We estimate that intermittent inpatient EEG misses 2–4 times more subclinical seizures than the 2HELPS2B-determined acceptable 5% seizure miss-rate threshold for cEEG. Future research is needed to determine the impact of potentially missed seizures on clinical care.