Abstracts

Hyperventilation (HV) is used during routine EEG to activate epileptiform abnormalities (EA) but was limited in use during the COVID-19 pandemic. Here we evaluated the effect of HV on EEG-predictors of epilepsy, comparing findings in the standard EEG versus HV-limited protocol. We evaluated other activating procedures and parameters (sleep, sleep deprivation, photic stimulation, study duration) to place HV in context of other factors thought to precipitate EA. We hypothesized that HV would not significantly impact the sensitivity of routine EEG in a population of pediatric and adult patients.

We conducted a retrospective study of routine EEG completed at UVA between March 2018 – March 2022. Subjects were >4 years old with at least fair HV effort; control subjects did not HV. The primary outcome variable was EA, defined as clinical-electrographic or electrographic seizures, spikes or sharp waves, or temporal or occipital intermittent rhythmic delta activity. We screened the relationships of dependent variables against EA status with Fisher’s exact tests or chi-square tests. We evaluated logistic regression models with the dependent outcome of EA (present or absent) against independent variables of HV, EEG duration category (≤ 39, >39 minutes), photic stimulation, sleep deprivation, and sleep as well as patient-related variables of age group (< 12, 12-17, >17 years), sex, and number of daily antiseizure medications (ASM group, ≤1 or >1). We displayed the resulting odd-ratio of independent variables with a Forest plot. We evaluated the subset of patients with serial EEGs with HV performed during one study (usually in the pre-COVID period) and HV not performed during another (usually in the post-COVID HV restriction period). We compared the appearance of EA with and without HV within individuals via McNemar’s test.

We reviewed 659 EEGs from unique patients (806 EEGs reviewed). Of these, 180 had EA (27%). Once other variables were accounted for with logistic regression (in rank order of odd ratios), ASM >1, sleep deprivation, longer EEG duration, and HV all had significant positive odds ratios on the aggregate prevalence of EA (Figure 1). Older age decreased the odds of capturing EA. Of the total sample of 659 patients, 64 (9.7%) underwent HV for one EEG and did not for a second. Among this group, 34 (53%) had EA with HV, and 26 (41%) had EA without HV. Although 12 patients with no EA without HV had EA captured with an HV study, four without EA during an HV study had EA on a non-HV study, for a net change of eight patients. This change was not significant (McNemar’s test).

After accounting for age and other patient-related factors, multiple ASMs, sleep deprivation, longer EEG duration, and HV favored the presence of EA. Adults are less likely to have EA. Patients on multiple ASMs are more likely to have EA. In a subset of patients with EEGs with and without HV, HV did not significantly favor EA activation in that recording. Thus, while HV favors EA appearance its effect is limited.