Abstracts

Rationale: Hyperventilation is a standard activating procedure used in routine EEG but was limited in use due to the COVID pandemic. A prior study has shown decreased sensitivity of EEG when omitting HV in juvenile myoclonic epilepsy patients during COVID (Hatano 2022). We sought to investigate the potential decrease in sensitivity of EEG more broadly in both pediatric and adult patients. We reviewed routine EEG reports from before and during the COVID pandemic and assessed the impact of omitting this technique on the sensitivity of routine EEG. We hypothesized that HV did not significantly impact the sensitivity of EEG in a broad population.

Methods: We conducted a retrospective study of patients aged 4 years and older at the University of Virginia Health System who underwent routine EEGs from 2018-2022. This study was approved by the University of Virginia IRB. Data was extracted EEG reports and supplemented by EEG review. The primary outcome variable was the presence or absence of epileptiform abnormalities (EAs) defined as either 1) clinical-electrographic or electrographic seizures, 2) spikes or sharp waves, or 3) occipital or temporal rhythmic delta activity. The primary independent variable was adequate HV (present = fair or better effort by the technologist; absent = poor or no HV performed). Other independent variables were sex, age, sleep (presence or absence), EEG duration (defined as 21 minutes or less versus greater). We screened independent variables against the presence of absence of EAs with univariate T tests and chi squared tests. We calculated the odds ratios of independent variables against IED outcome with stepwise binary logistic regression models.

Results: Of 206 patients, 73 had EAs (35.4%). HV was not associated with a higher proportion of the presence of EAs. (Table 1). Univariate screening revealed a significant association with EA presence with the presence of sleep, duration greater than 21 minutes, and the total number of ASMs reported at time of EEG. Patient age and sex were not associated with EAs. HV was not a significant variable in any logistic regression model. Longer EEG duration (OR[95% confidence limit] 4.69[1.21-18.17], p=0.025) and ASM number (2. 53[1.31-4.89], p=0.006) were significantly associated with presence of AEs taking into account all other independent variables (Figure 1).

Conclusions: The yield of EAs in EEG is not enhanced by HV even when adjusting for patient and EEG-based variables. Whereas rote use of HV protocols may not be clinically useful, longer EEG durations clearly may improve sensitivity of EEG in support of diagnosis of epilepsy.

Funding: None