Abstracts

Rationale: Patients suffering from nonconvulsive seizures in the intensive care unit (ICU) currently experience significant delays in diagnosis and treatment with current electroencephalography (EEG) technology. The Brain Stethoscope device (BSD) is a novel 10-lead EEG device that can be placed on a patient in minutes and converts raw EEG waveform data to sound, allowing EEG interpretation by any clinician. We hypothesized that use of the BSD on encephalopathic patients in the ICU would result in significantly decreased time to acquire EEG data and improved diagnosis of and decision to treat nonconvulsive seizures.   Methods: IRB approval was obtained.  Patients admitted to any ICU at Stanford Hospital were enrolled from 7:00 AM to 5:00 PM if they were encephalopathic (GCS < 12) and if the treating team’s suspicion for nonconvulsive seizures warranted continuous EEG (cEEG) monitoring.  Once the cEEG order was placed, consent was obtained, the BSD was set up by the treating physician, and sonification was performed on left and right hemispheres independently for 30 seconds each. The physician rated suspicion for seizure (1-5; 1=low, 5=high) and decision to treat with AEDs (yes/no/not sure) before and after sonification using a standard form. All BSD EEG waveform data was also evaluated by an independent epileptologist blinded to clinical data to assess accuracy of sonification. Outcomes were a) the difference in time between cEEG order to start of BSD and cEEG recording, b) the change in suspicion for seizure pre- and post sonification (% rated as 1-low suspicion), and c) the change in decision to treat (treat vs no/not sure).   Results: Thirty-two adult patients with a mean age of 60 + 18 years and median GCS of 6 (IQR 4-8.5) were enrolled. Median time from cEEG order to start of monitoring was significantly less for BSD over continuous EEG monitoring (0.36 hours (IQR 0.21-0.73) vs 2.57 hours (IQR 1.63-3.95)), with median difference of 1.77 hours (IQR 1.17-2.75, min 0.48, max 8.52). One patient had an electrographic seizure during sonification, and this was accurately identified (change in suspicion increased from 1 to 5; decision to treat changed from no treatment to treat). Post-sonification, more physicians reported low suspicion (rated as 1) for seizure activity (66%, vs. 38% pre-sonification, p=0.03). Treatment decision changed in 14/32 (44%) patients after sonification (figure 1), and this was in the appropriate direction 86% of the time as compared to the independent epileptologist’s EEG waveform interpretation during sonfication.  Sonification also decreased inappropriate decision to treat from 28% to 6% (p=0.02).  Negative predictive value of BSD sonfication was 1.0 (95% CI, 0.84-1.0).     Conclusions: Use of the Brain Stethoscope on ICU patients suspected of having nonconvulsive seizures resulted in faster time to EEG acquistion, accurate identification of non-convulsive seizure by sonification, and appropriate change in physician suspicion of seizure and treatment decision. The Brain Stethoscope has the potential to greatly improve timely diagnosis and management of seizures in critically ill patients. Funding: Funding provided by Ceribell, Inc.