Rapid-Response EEG in Children With Suspected Status Epilepticus

The primary aim of this study is to investigate by how much faster EEG results can be obtained using the Ceribell® Rapid Response EEG (RR-EEG) device (CERIBELL, INC., Sunnyvale, CA) as compared to conventional continuous 20-channel EEG (cEEG). The study is conducted in a single urban academic pediatric hospital 26-bed ICU as a retrospective cohort study by chart review of pediatric patients who either underwent conventional continuous EEG (cEEG) or rapid response EEG (RR-EEG) as their first EEG study within 24 hours of admission. The indication for an EEG study was determined by the treating ICU team based on the suspicion of status epilepticus and the choice of EEG modality was at the primary physician's discretion. This retrospective study was approved with a waiver for consent by the Institutional Review Board at Albert Einstein College of Medicine (IRB # 2021-13666).

Subjects:

Pediatric patients (0-21 years) primarily admitted to the pediatric ICU (PICU) with a neurological diagnosis, diminished level of consciousness, and established or suspected status epilepticus are included. Exclusion criteria are rapid neurological improvement, traumatic brain injury and incomplete (<1 hour recorded) EEG data.

The investigators will query all studies on Natus® (Middleton, WI) NeuroWorks® (cEEG) and the online Ceribell® portal (RR-EEG) to review continuous EEG studies between January and December 2020 to capture patients admitted to PICU with complete EEG data. Through chart review the investigators verified that EEG studies were started within 24 hours of admission. If the numbers obtained aren't sufficient (see statistics section) the data collection will be expanded to include the years 2021-2022.

The primary outcome is times from EEG order to first EEG report. The time of EEG order will be used as a surrogate for the treatment team's decision to obtain an EEG. Data collection also includes covariables that could potentially affect the primary outcome (potential confounders) based on patient factors (age, sex, race, head circumference, known epilepsy, baseline pediatric Glasgow Outcomes Score Extended [GOS-E Peds] and comorbidities), clinical factors (witnessed clinical seizures, decreased level of consciousness, Glasgow Coma Scores, admission diagnosis, neurological diagnosis and Paediatric Index of Mortality score Version 2 [PIM-2]) and management factors (intubation, imaging obtained, procedures performed in the ICU, rate of admission during normal business hours). These data will be collected to address potential bias in this study.

For secondary outcomes the investigators will collect times from order to EEG start ("deployment time"), EEG start to EEG report and EEG findings (electrographic seizures, epileptiform activity and EEG background findings). The investigators also collect data on anti-seizure medication management and categorized anti-seizure medication management as escalation, maintenance, and de-escalation through chart review by an investigator blinded to the EEG results. Addition or up-titration of the doses of anti-seizure medications (lorazepam, midazolam, levetiracetam, Propofol, valproate, fosphenytoin or baseline anti-seizure medications) will be considered escalation of therapy, no change will be considered maintenance, and reduction of the dose or discontinuation of medications will be considered de-escalation. ICU and hospital length of stay, GOS-E Peds and discharge disposition (home, rehab or nursing facility) will also be collected. All data sources for both groups are obtained by electronic medical record chart review by research staff not primarily involved in patient care.

RR-EEG Device, Training and System:

Our PICU started using the Ceribell® RR-EEG tool for detecting electrographic seizures in children suspected to be in non-convulsive status epilepticus in 2020. This is in parallel to cEEG that is still used in this ICU. The Ceribell® RR-EEG device consists of a 10-electrode headband and records a total of 8 EEG channels approximating the temporal chains in the 10-20 system (the parasagittal chains are omitted). The EEG study is wirelessly uploaded to a HIPAA compliant online portal that can be accessed remotely.

Pediatric intensive care physicians determined the need for EEG monitoring, ordered and initiated a Ceribell® RR-EEG study, or ordered a conventional EEG (cEEG) alone. Each participating physician completed a training session on the use of rapid response-EEG, which included watching a video describing its setup (20 min) followed by a qualification test and a hands-on session (45 min) that included a live demonstration of rapid response-EEG setup. RR-EEG studies were followed by a conventional 20-channel EEG when the resource was available at the discretion of the primary team and the neurology consultants.

Definitions:

The definition for seizure description followed the recommendations of the International League Against Epilepsy (ILAE) position paper. EEG reports were generated using templates following the American Clinical Neurology Society's Standardized Critical Care EEG Terminology. These could be modified at the discretion of the reading physician. The report narrative will be used to categorize findings into presence or absence of seizures, potential epileptiform activity (spikes, sharps, periodic patterns), and the degree of cerebral dysfunction interpreted as representing normal, mild, moderate or severe dysfunction (based on EEG background descriptions such as baseline frequency, anterior-posterior gradient, posterior dominant rhythm and reactivity).

Business day are defined as 9 AM to 5 PM Mondays to Fridays except for holidays. All other times are defined as "weekends or nights".

The Pediatric Glasgow Outcomes Scale - Extended (GOS-E Peds - a scale from 1 [best] to 8 [worst]) will be used to categorize clinical outcomes. This tool was preferred over the commonly cited Pediatric Cerebral Performance Category (PCPC) because of its higher granularity (15, 16). The GOS-E Peds classification will be graded retrospectively based on data from discharge summaries. Good outcome is defined as GOS-E Peds ≤ 3. Disposition (home, acute rehabilitation, long-term nursing facility or death) will be used as a surrogate for outcomes on hospital discharge.

Statistical Analysis:

Based on prior publications in adult ICU populations, it appeared feasible to obtain a preliminary EEG report in under 2 hours, using the Ceribell® RR-EEG device. For a clinically significant reduction in report time by 2 hours (120 min), with a standard deviation of 2 hours (120 min), the investigators estimate a sample size of a total n = 60 with following parameters: alpha = 0.025, power = 0.9, RR-EEG proportion of 33%, i.e., 20 RR-EEG subjects and 40 cEEG subjects.

Descriptive analysis will be performed by summarizing data as medians and interquartile ranges (IQR) for continuous variables and frequencies and percentages for categorical variables. Variables will be compared using Fisher exact tests and Wilcoxon rank sum tests for categorical and continuous variables, respectively. Multivariable linear regression analysis will be employed to evaluate RR-EEG vs. cEEG on the primary outcome, while adjusting for potential confounders. Factors which are significantly different between both groups will be included in the multivariable model. Assuming that the outcome will not normally distributed, a log transformation will be applied. Results will be converted back into anti-log scales and reported as the ratio of the outcome, which can be represented as a percent change in the outcome for ease of interpretation. P-value < 0.05 will be considered statistically significant.