Nonconvulsive Seizure Identification on Quantitative EEG by Neuro ICU Nurses

PROTOCOL TITLE A prospective trial of nonconvulsive electrographic seizure detection by
nurses in the adult Neuro ICU using a panel of quantitative EEG trends

PURPOSE OF STUDY The primary objective of this prospective study is to determine the
sensitivity and specificity of quantitative EEG (qEEG) trends in the identification of
recurrent nonconvulsive electrographic seizures by non-neurophysiologists in adult patients
admitted to the Duke Neuro Intensive Care Unit (Neuro ICU). Our hypothesis is that bedside
Neuro ICU nurses using a panel of qEEG trends can identify recurrent nonconvulsive seizures
(NCS) with adequate sensitivity and specificity.

Secondary objectives include identifying seizure and EEG characteristics that make NCS more
or less likely to be detected on qEEG. These characteristics include average seizure
duration, spatial extent of the seizure (focal, hemispheric or bilateral/generalized) and the
EEG background (periodic vs. non-periodic and epileptiform vs. non-epileptiform).

PRINCIPLE INVESTIGATOR Dr. Saurabh R. Sinha MD, will be the principal investigator (PI) on
this project and is responsible for the study design, generation and analysis of data,
presentation of reports, and protection of human subjects in the performance of the clinical
investigation. His contact information is provided below.

Saurabh R. Sinha, M.D. 295 Hanes House, 330 Trent Drive. Box 102350 Duke University Medical
Center. Durham, NC 27710. Phone: 919-684-8485. Fax: 919-684-8955. Email:
saurabh.sinha@duke.edu

BACKGROUND AND SIGNIFICANCE

Due to the increased awareness of nonconvulsive seizures (NCS) and nonconvulsive status
epilepticus (NCSE) in patients admitted to the NICU, the utilization of continuous EEG (cEEG)
monitoring in the NICU is rising. qEEG is a potential tool that can aid in the interpretation
of large volumes of EEG data. qEEG is the application of mathematical and analytical
techniques to characterize and analyze EEG signals. Several hours of data can be displayed on
a single screen in several graphical panels. Routine evaluation of EEG data consists of
manual inspection of each page of EEG data and can be quite time-consuming. Since continuous
review of EEG data is not possible, EEG data is always relayed to the primary clinical team
in a post-hoc manner which is not ideal. Theoretically, qEEG graphical displays can provide a
simplified view of complex EEG data that could possibly be used by non-neurophysiologists,
such as Neuro ICU nurses.

There have been no studies in the adult or pediatric population that have prospectively
evaluated the use of qEEG by non-neurophysiologists to identify recurrent NCS. Given that our
retrospective data show that neurophysiologists and non-neurophysiologists have similar
sensitivity and specificity in identifying the presence of NCS on qEEG panels, our hypothesis
is that Neuro-ICU nurses will have an acceptable sensitivity and specificity in prospectively
identifying recurrent NCS.

DESIGN AND PROCEDURES

This is an observational, prospective evaluation of Neuro ICU nurses' ability to detect NCS
via qEEG running at the bedside. The study will not affect clinical management decisions or
patient care. It will utilize DUHS Neuro ICU nurses that will serve as qEEG readers.

The goal of this study is to compare the accuracy of Neuro-ICU nurses' ability to detect NCS
on bedside qEEG and post hoc review of the corresponding raw EEG by blinded, board-certified
neurophysiologists. The neurophysiologists are blinded to the Neuro ICU nurses prospective
seizure identification.

We plan to enroll a sample of 40 nurses and 30 adult patients who are already on continuous
video electroencephalogram (EEG) monitoring during their admission to the Duke Neuroscience
ICU. In this prospective study we will include adults who are clinically suspected to have
NCS, defined by the fact that the treatment team physician has requested cEEG monitoring. The
treating physician will make the decision to initiate cEEG monitoring and determine the
duration of monitoring. We will exclude subjects with convulsive seizures without concern for
NCS.

All eligible subjects will be consented for enrollment after the clinical team has decided
that the patient will be placed on cEEG monitoring. Subsequently, the neurophysiology fellow
will review the raw EEG intermittently as this is the standard of care. Once the
neurophysiology fellow has detected NCS, the research team will be contacted by either the
clinical Neuro ICU team or the neurophysiology team if the patient is deemed appropriate to
continue participation in the study.

Nurses will be consented for their participation in the trial as it involves their ability to
detect NCS with the use of bedside qEEG trends. Neuro ICU nurses who consent will receive
training on qEEG monitoring by the research team via a 15 minute online PowerPoint
presentation. In addition, a binder containing printouts of the training PowerPoint
presentation will be attached to each EEG machine.

A single printout of a qEEG panel, containing the sentinel seizure captured on cEEG
monitoring will be displayed next to the cEEG machine. The NCS will be clearly marked by the
research team. All patients undergoing cEEG monitoring at Duke have qEEG running
simultaneously in addition to the raw EEG. The program that creates the qEEG display is Magic
Marker developed by Persyt, Inc. Neuro ICU nurses will be asked to check the qEEG screen
hourly and record the number of seizures (similar in appearance to the sentinel seizure qEEG
print out) seen on the qEEG display. A 24 hour log will be attached to the EEG machine. If a
patient and a nurse are enrolled in the study, but the second nurse on the later shift has
not agreed to participate in the trial then we will either end the study or ask the second
nurse if he/she would like to participate. There are nurses from the float pool that may be
interested, and if so, we will provide a one-on-one training session at the beginning of
their shift and consent them. At one hour intervals the nurse will mark one of the following
options: no seizures, 1-2 seizures, 3-5 seizures, 6-10 seizures or >10 seizures. The absolute
times that the nurse should evaluate the qEEG display will be determined by the study team at
the time of enrollment and clearly marked on the log sheet. The standard of practice of the
neurophysiology fellow independently reviewing the raw EEG intermittently will continue
unchanged. Treatment of recurrent NCS will be at the discretion of the Neuro ICU team in
conjunction with the neurophysiology team. The nurse's review of the qEEG will have no impact
on patient care or treatment decisions.

The nurse will review the subject's qEEG for the duration of their shift (12 hours) or for
only part of their shift. The total duration of the study will be for 24 hours or when the
treating physician discontinues continuous EEG monitoring, whichever is shorter. If the first
nurse participates in the study, but the nurse on the second shift has not agreed to
participate, then the study will stopped.

After completing enrollment, the qEEG data and raw EEG data will be de-identified by study
author (CW). Board-certified neurophysiologists and study authors (SS and CS) will review the
raw EEG data independently to identify seizures. Electrographic seizures will be identified
using published criteria [15]. In the case of discrepancy in the identification of seizures
the two authors will review the studies together and establish a consensus. Each hour of raw
EEG will be placed into the follow categories: no seizures, 1-2 seizure, 3-5 seizures, 6-10
seizures or >10 seizures. In addition, the average seizure duration, the spatial extent of
each seizures (focal, hemispheric or generalized/bilateral), and primary background EEG
abnormalities will be determined by the author and board-certified neurophysiologist, CS,
from the raw EEG.

SELECTION OF SUBJECTS

Neuroscience ICU Nurses:

All Neuroscience ICU nurses will be asked to participate.

Subjects:

The responsible treating team in the Adult Neuro ICU will identify subjects after the
treating physician has decided to initiate cEEG monitoring. No restrictions are made with
regard to gender, race, religion or underlying cause for hospitalization, including prior
history of seizure disorder.

SUBJECT RECRUITMENT AND COMPENSATION Appropriate subjects will be identified by the treating
team in the Adult Neuro ICU. It will be at the sole discretion of the treating team in the
Adult Neuro ICU or the clinical neurophysiology team to notify a research coordinator to
approach the patient for recruitment into the study. Subject demographics will mirror that of
the patient demographics of the Duke Adult Neuro ICU. No compensation will be provided.

CONSENT PROCESS

Neuroscience ICU Nurses:

The study will be introduced to all Neuro ICU nurses via an email request to participate in
the study. The consent form will be reviewed with all nurses willing to participate in the
trial. The study physicians will review the consent forms individually with each nurse in the
Duke Medical Pavilion conference room and answer any questions individually at that time.
Those that agree to participate in the study and sign the consent form will attend a brief,
15 minute training session on qEEG monitoring presented by the study authors (SS, CS and CW).
Their participation is completely voluntary and their recorded responses of qEEG
interpretation will be used as part of the project data. No identifying information will be
recorded. It will be made explicitly clear that there will be no penalty for not
participating and that no incentives will be provided.

Subjects and Subjects Legal Representative:

The study will be introduced to the patient or Legally Authorized Representative (LAR) by a
member of the clinical team. If the patient or LAR is willing to participate, the study team
will approach them. Informed consent will be obtained by the PI, Study Coordinator, or other
Key personnel for all subjects in the study. Consent will be obtained from the patient or
LAR, as defined by Duke guidelines, if the patient is unable to provide consent. The patient
will be assessed for impaired decision-making at the time of consent. All patients requiring
cEEG monitoring will require involvement of a neurologist who will determine the ability to
consent if that determination has not already been made either by the PI or other
investigators on the trial. All investigator personnel on this trial are neurologists.
Consent will be obtained in a quiet, private location to protect the privacy of the
prospective participant. Potential participants will be identified by bedside evaluation at
DUHS Neuro ICU. Once the participant is identified, the patient and/or family will be
notified of the opportunity to participate in the project, and the purpose of the study will
be reviewed in addition to all of the risks and benefits of participation. The entire consent
form will be reviewed and all questions will be addressed. It will be made explicitly clear
to all participants and LARs that no additional procedures will be performed as part of this
trial and that no incentives will be provided.

SUBJECT'S CAPACITY TO GIVE LEGALLY EFFECTIVE CONSENT Many subjects will not have the capacity
to provide consent given the nature of the study and acuity of the disease. Consent will be
obtained from the subject's legal representative. Periodic assessment of capacity will be
assessed by the research coordinator, and if deemed to have capacity the subject will be
presented with the opportunity to be consented.

STUDY INTERVENTIONS The standard of care will continue unchanged, and this consists of
intermittent raw EEG review by the neurophysiology team. No interventions will come to the
participating subject based on qEEG data review by the Neuro ICU nurses, but rather will
continue to rely on the current standard of care with the existing infrastructure using raw
EEG. All decisions to treat NCS will be at the sole discretion of the treating team and
responsible care physicians. The duration of continuous EEG monitoring will be determined
solely by the treating team or responsible care physician.

RISK/BENEFIT ASSESSMENT

Subjects and Subjects Legal Representative:

There are no physical, psychological, legal, or economic risks identified. Social risks
include the decision to choose to participate or not participate. Cognitively impaired adults
may be recruited as subjects given the acuity and sequela of NCS. Special precautions include
the deferring to the subjects legal representative for consent in this situation. Should the
subject or subjects LAR choose not to participate there will be no repercussions and medical
care will continue per the treating team. There will be no direct benefit to the patients
enrolled in the study.

Neuroscience ICU Nurses:

There are no physical or psychological risks identified. The nurse's decision to participate
or not will not have repercussions. The decision to participate will have no ramifications on
employment status with Duke University Hospital (job classification, pay, etc.). The
neuroscience ICU nurses will benefit from the study by receiving training on qEEG
interpretation.

COSTS TO THE SUBJECTS

Subjects and Subjects Legal Representative:

There will be no additional costs that the patient's will incur as a result of participation.

Neuroscience ICU Nurses:

There will be no costs that the Neuro ICU nurses will incur as a result of participation.

DATA ANALYSIS AND STATISTICAL CONSIDERATIONS We will determine the sensitivity and
specificity of the use of qEEG trends by Neuro ICU nurses in identifying recurrent
nonconvulsive seizures in critically ill patients with acute or chronic neurologic injuries.
This will be calculated by comparing the number of suspected seizures marked by Neuro ICU
nurses via qEEG displays to the number of seizures identified by the gold standard analysis
of the raw EEG.

Standard test characteristics (sensitivity, specificity, positive predictive value and
negative predictive value, positive likelihood ratio and negative likelihood ratio) will be
calculated for the ability to detect the presence or absence of seizures. Contingency
analysis will be used to make several comparisons of the mean diagnostic accuracy of the
Neuro ICU nurses: 1. short (< 25th percentile) vs. long (> 75th percentile) seizures; 2.
spatial extent of seizures (generalized vs. hemispheric vs. focal); 3. Background EEG pattern
(epileptiform vs. non-epileptiform and periodic vs. non-periodic). The chi-square statistic
will be calculated for each these analyses. For all calculations, a p-value < 0.05 will be
considered statistically significant.

DATA AND SAFETY MONITORING Raw and qEEG data will be obtained and stored (as it is
currently). The qEEG and raw EEG data will be anonymized after the subject has completed
participation in the study. Although all EEG data in the Neuro ICU is recorded with video for
clinical purposes, no video will be retained during this project. The anonymized data will be
used solely by research investigators.

Standard clinical practice will not be altered during this study. There should be no impact
on patient care.

PRIVACY, DATA STORAGE AND CONFIDENTIALITY

Data to be collected will include the digital EEG portion of our cEEG monitoring records and
their corresponding qEEG panels. Although all EEG data in the Neuro ICU is recorded with
video for clinical purposes, no video will be retained during this project. After extracting
the relevant data from the complete cEEG file, all identifying information will be removed
and the EEG will be numerically indexed in the data set. All cEEG and qEEG data is currently
collected locally (bedside PC) on the monitoring system and automatically copied to the
central server for remote review. Once the data is collected, the files are centrally
reviewed, and at that point can be copied to the project computer for indexing, extraction of
segments and de-identification of the data. All PHI data will be removed leaving no record of
the origin of the data. Clinical data will be stored on the Duke IT maintained network
server. The de-identified cEEG and qEEG data will be stored on the Isilon storage server
behind Duke's firewall. The de-identified cEEG or qEEG data will not be shared with any
parties outside Duke.

Clinical data will be collected from the Duke electronic medical record system. This includes
age, race, gender, primary diagnosis, and reason of cEEG monitoring. As part of the study
protocol, no PHI will be recorded at any time. PHI will only be reviewed to extract relevant
study information, such as patient demographics and official interpretation of the cEEG by
the neurophysiology team. This review process will occur less than 24 hours after study
initiation. After the relevant information is extracted, the demographic data, cEEG data and
qEEG data will be anonymized.

Research data will be stored on a Duke IT maintained network server. Confidentiality will be
ensured as no PHI or video data will be retained on this server. The de-identified cEEG and
qEEG data will be used solely by research investigators.

Inclusion Criteria:

1. Age of 18 years or older at time of consent

2. Patients undergoing cEEG monitoring to evaluate for the presence of NCS

3. Admission or pending transfer to the adult Neuro ICU

4. Informed consent by patient or patient's legally authorized representative (LAR)

5. One or more NCS identified on cEEG monitoring, confirmed by the clinical
neurophysiology team

Exclusion Criteria:

1. Anticipated discharge from NICU < 24 hours

2. Anticipated duration of cEEG monitoring < 24 hours

3. Suspected diagnosis of brain death

4. Open head wound or bandage preventing application of majority of EEG electrodes

5. Treating physician deems not appropriate for subject participation

6. Seizures lasting more than 1 hour in duration

7. Convulsive seizures without concern for NCS