Clinical Neurophysiology 123 (2012) 910-917
Contents lists available at SciVerse ScienceDirect
Clinical Neurophysiology
j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / c l i n p h
Emergent EEG in the emergency department in patients with altered mental states
Wendy C. Ziai a,b,, Dan Schlattman c, Rafael Llinas a, Santosh Venkatesha c, Melvin Truesdale c,
Anastasia Schevchenko a, Peter W. Kaplan a
a Department of Neurology, Johns Hopkins University School of Medicine, Baltimore, MD, United States
b Department of Anesthesiology and Critical Care Medicine, Johns Hopkins University School of Medicine, Baltimore, MD, United States
c Infinite Biomedical Technologies (IBT), Baltimore, MD, United States
See Editorial, page 855
a r t i c l e
i n f o
h i g h l i g h t s
Article history:
Availability of a rapid, standard full montage EEG in the emergency department is a feasible goal.
Available online 5 October 2011
Specific presentations of altered mental status offer the best diagnostic benefit of EEG in the emergency
department.
Keywords:
Abbreviated 5 min full-montage EEG presents adequate reliability which should improve acceptance
Electroencephalogram
and use of EEG in the emergency department.
Consciousness disorders
Emergency service/hospital
Seizures
a b s t r a c t
Objective: To evaluate whether EEG performed within 30 min of referral by an ED physician helps estab-
lish diagnosis and/or changes management and in which clinical setting.
Methods: Single-center prospective cohort intervention study 1 day/week, of sequentially referred adult
patients with clinical seizures or altered mental status (AMS). Standard EEGs were performed by an EEG
technician using a commercially available cap, interpreted by an epileptologist, immediately reported to
the ED physician and a utility survey completed. Quality and interpretation of 20 min EEGs was compared
to pre-specified 5 min segments of each EEG using the kappa coefficient.
Results: Over 1 year, 82 patients underwent ED EEG. Tonic clonic seizure activity had occurred in 33%.
Mean time for EEG setup was 13.1 6.2 min. EEG assisted the diagnosis in 51%, changed ED management
in 4% and would be ordered again if EEG was available in 46%. Positive utility of EEG was significantly asso-
ciated with toxicologic, psychiatric and endocrine/metabolic causes of AMS vs. other causes (p < 0.001)
and sudden onset AMS (p = 0.007). Independent predictors of whether ED EEG would be ordered if avail-
able were witnessed seizures (p = 0.01), no prior head trauma (p = 0.001) and survey respondent being a
physician assistant (vs. MD) (p = 0.02). The 5 (vs. 20) min EEG presented good agreement on waveform
shape/amplitude (kappa = 0.78), artifact (kappa = 0.75) and interpretation categories (all kappa levels
P0.70).
Conclusions: Rapid availability of standard full-montage EEG in the ED is feasible and helps establish a
diagnosis in about half of AMS patients, but rarely changes management. An abbreviated 5 min full-mon-
tage EEG presents adequate reliability which may improve use in the ED.
Significance: Specific presentations of AMS offer the best diagnostic benefit for EEG in the ED.
O 2011 International Federation of Clinical Neurophysiology. Published by Elsevier Ireland Ltd. All rights
reserved.
1. Introduction
2002; Johnson et al., 1994; Erkinjuntti et al., 1986; Wofford et al.,
1996). Seizure is the presenting symptom in 1.2-3.5% of all emer-
Up to 10% of patients in the emergency department (ED) present
gency department evaluations (Nandhagopal, 2006; Huff et al.,
with seizures or altered mental status (AMS) (Wolfe and Brown,
2001; Meena et al., 2001), and between 5% and 17% of patients will
have seizures in the ED (Huff et al., 2001; Kanich et al., 2002). The
Corresponding author at: Johns Hopkins Hospital, Division of Neurosciences
most common causes of AMS in the ED are neurologic accounting
Critical Care, 600 N. Wolfe Street/Meyer 8-140, Baltimore, MD 21287, United States.
for 28% of discharge diagnoses in one study of 317 patients present-
Tel.: +1 410 614 6121; fax: +1 410 614 7903.
ing with AMS (Kanich et al., 2002). AMS is a significant challenge to
E-mail address: weziai@jhmi.edu (W.C. Ziai).
1388-2457/$36.00 O 2011 International Federation of Clinical Neurophysiology. Published by Elsevier Ireland Ltd. All rights reserved.
doi:10.1016/j.clinph.2011.07.053

---

W.C. Ziai et al. / Clinical Neurophysiology 123 (2012) 910-917
911
the emergency physician because it is a manifestation of a wide
sensors to change as the system is stretched and secured around
range of medical syndromes and does not indicate a specific diagno-
the patient's head, but remain in the correct 10-20 localizations.
sis (American College of Emergency Physicians, 1999). Clinical
EEG collection was at a low frequency filter (LFF) of 0.01 Hz, and
evaluation is difficult especially if a thorough history cannot be ob-
a high frequency filter (HFF) setting of 70 Hz. Ceegraph imped-
tained and the etiology is often not immediately obvious. Not sur-
ances obtained at the beginning of the recordings were set not to
prisingly rates of ED resource use, hospital admission and death
exceed 25 kX.
for this diagnosis are disproportionately high relative to the per-
EEGs were interpreted by a Board Certified Electroencephalog-
centage of the ED population this group represents (Kanich et al.,
rapher informed of the clinical scenario (EEG expert - PK,RL) with-
2002). The electroencephalogram (EEG), the most widely accepted
in a goal of 30 min. For interpretation, the EEG filter settings, paper
test of electrocerebral activity, is underused, or often not available
speed and sensitivity along with the 60 Hz notch filter were indi-
in the ED (Hoosmand and Maloney, 1980; Kaplan, 1999). Proposed
vidualized for optimal interpretation and subsequent display pur-
reasons for this underuse include lack of availability of a technolo-
poses in the figures provided. The EEG interpretation was
gist, difficulty and time required for signal acquisition (American
immediately conveyed to the ED physician usually in person or
Clinical Neurophysiology Society, 2006), lack of timely expert signal
by telephone (1 case) and a survey completed by the ED physician
interpretation (Quigg et al., 2001) and lack of awareness of utility of
or physician assistant (PA) about the usefulness of the EEG (utility
EEG in this setting.
survey).
Criteria for EEG in the ED have not been established. The pri-
mary objectives of this study were to determine: (1) whether stan-
2.2. EEG interpretation
dard full-montage EEG performed within 30 min of referral by an
ED physician helps to establish the diagnosis and/or changes ther-
EEGs were rated by an expert neurologist (PK or RL) on a scale
apeutic management in the ED; and (2) which clinical features and
from 1 to 5 (1: very poor; 2: poor; 3: satisfactory; 4: good; 5: excel-
suspected diagnoses are useful in selecting patients for ED EEG. We
lent) for (1) waveform shape and amplitude; and (2) overall
also assessed whether using a full montage, but abbreviated 5 min
appearance of artifact or noise in the signals. The ED EEG results
EEG with expert interpretation provides adequate screening reso-
were categorized as follows: (1) diffusely abnormal without elec-
lution and has comparable detection capability to standard ED EEG.
trographic seizures. This category included background slowing
and diffuse bisynchronous or multifocal interictal epileptiform dis-
charges that did not evolve into seizures. (2) Diffusely abnormal
2. Methods
without epileptiform discharges. (3) Focal abnormalities without
electrographic seizures. This category included focal interictal epi-
2.1. Design
leptiform discharges that did not evolve into seizures. (4) Focally
abnormal without epileptiform discharges. (5) Electrographic sei-
This was a single-center prospective cohort intervention study
zures. This category included generalized and focal seizures. (6)
1 day/week, of sequentially referred adult patients presenting to
Normal EEG. (7) Unable to interpret. More than one category could
the emergency department with witnessed clinical seizure activity,
be identified for a single EEG recording if the EEG contained mixed
either focal or generalized or AMS. Altered mental status was
features.
defined as a new onset change in the level of arousal and quality
To determine whether using an abbreviated 5 min EEG with ex-
of awareness that could present as a state of lethargy, aphasia,
pert interpretation provides adequate screening resolution and has
psychosis, obtundation or coma that was not induced by
comparable diagnostic detection capability to standard ED EEG, the
medications given by a medical provider. The condition could be
EEG experts graded a 5 min segment (minutes 10-15 of the 20 min
improving or ongoing at time of enrolment.
ED EEG on each patient) retrospectively remaining blinded to clin-
The study was conducted at the Johns Hopkins University Bay-
ical information and the original EEG result. Each EEG expert only
view Medical Center, a tertiary care ED in Baltimore, MD for one
graded 5 min EEG segments of the same EEGs they had already
year (July 2009 through June 2010). This 750-bed hospital has a
graded in the ED.
referral Neurology/Neurosurgery Program and the ED manages
approximately 58,000 patients annually. It is staffed with full-time
2.3. Statistical analysis
emergency physicians and physician assistants and has an emer-
gency medicine residency program. Over the study period, the
The primary endpoints were the proportion of patients with
number of patients admitted with altered mental status or seizures
AMS in whom ED EEG (1) helped to establish a diagnosis, (2) would
were 388 and 64, respectively. The study was approved by the
be ordered again based on perceived usefulness by the ED
Institutional Review Board. Family members or patients if deemed
practitioner, and (3) changed therapeutic management. Univariate
clinically competent provided written, informed consent for all
analysis of factors associated with (1) and (2) was performed using
subjects.
Student's t-test for continuous variables with normal distributions
Patients meeting inclusion criteria were referred by the ED phy-
and Chi-squared or Fisher's exact test for analysis of categorical
sician to the on-site EEG technician without neurologic consulta-
data as appropriate. Multivariate analyses were performed with lo-
tion. Imaging and laboratory tests were prioritized by the ED
gistic regression models of associations of patient and EEG charac-
physician and were not used to decide whether a patient was eli-
teristics of interest (p < 0.1 in the univariate analysis) with whether
gible for the study.
EEG helped in establishing a diagnosis and whether an EEG would
Standard EEGs consisted of a minimum 20 min recording with a
be ordered by the ED practitioner on each particular patient if rou-
16-channel portable digital EEG machine (Bio-Logic) using 20
tinely available. Statistical significance was assigned for p 6 0.05.
hydrogel electrodes placed by the EEG technician according to
Data are presented as median interquartile range (iqr) or
the International 10-20 system. A commercially available FDA-
mean SD, unless otherwise indicated.
approved cap (Hydrodot EzeNet, HydroDot Inc., Westford, MA)
EEG signal quality and interpretation were compared between
was used that comes in multiple sizes, and incorporates disposable
full-montage 20 min and full-montage 5 min EEG using a Cohen's
hydrogel EEG electrodes, eliminating the need for skin preparation.
kappa coefficient for each EEG signal quality category and for each
The cap enables accurate positioning of sensors by fixing their loca-
EEG interpretation category (Cohen, 1960). This statistic adjusts for
tions on elastic bands. These allow the physical distance between
agreement occurring by chance. P(a) in Eq. (1) is the probability of

---

912
W.C. Ziai et al. / Clinical Neurophysiology 123 (2012) 910-917
Table 1
Table 3
Demographics of Patients with Altered Mental Status (N = 81).
EEG data acquisition times.
Demographics
No. (%) or median
Time interval
Time (mean SD) (min)
(interquartile range)
Mean time for EEG setup
13.1 6.2
Age (year)
57 (29)
Start of EEG setup to EEG transfer to EEG expert
55.0 25.7
Male
38 (46.9)
EEG complete to EEG expert arrival
20.4 25.3
Previous history of epilepsy
18 (22.2)
EEG expert arrival to report to ED staff
4.3 2.7
Use of chronic anticonvulsant therapy
17 (21.0)
Sub-therapeutic serum level of
8 (9.9)
anticonvulsant*
History of traumatic brain injury
Table 4
Acute
7 (8.6)
EEG interpretation of 20 min and 5 min ED EEG.
Remote
3 (3.7)
EEG interpretation
20 min EEG No.
5 min EEG No.
Previous history of cerebrovascular disease
18 (22.2)
(%) (N = 82)
(%) (N = 82)
(stroke, ICH, SAH)
Pre-existing diagnosis of dementing illness
10 (12.4)
Diffusely abnormal. May have
4 (4.9)
8 (9.8)
Current history of heavy ETOH use
7 (8.6)
epileptiform discharges. No seizures.
Use of sedatives, anxiolytics or narcotic
17 (21.0)
Diffusely abnormal. No epileptiform
22 (26.8)
2 (2.4)
analgesics
discharges
Current history of psychiatric illness
15 (18.5)
Focal abnormalities, May have
10 (12.2)
2 (2.4)
Clinical evidence of infection
10 (12.4)
epileptiform discharges. No seizures
Focal abnormalities. No epileptiform
4 (4.9)
0
Abbreviations: ICH: intracerebral hemorrhage; SAH: subarachnoid hemorrhage;
discharges
ETOH: ethanol.
Electrographic seizures
1 (1.2)
0
* If used for anticonvulsant properties.
Normal EEG
32 (39.0)
44 (53.7)
Unable to interpret
9 (11.0)
26 (31.7)
Table 2
Presenting characteristics of Patients with Altered Mental Status (N = 81).
Table 5
Presenting characteristic
No. (%)
EEG utility survey results.
Symptom onset before ED arrival
Survey Field
No. (%)
<6 h
59 (72.8)
6-24 h
8 (9.9)
Participants
>24 h
14(17.3)
ED MD
47 (60.3)
ED PA
31 (39.7)
Symptom onset
Gradual
22 (27.2)
Pre-EEG impression
Sudden
48 (59.3)
Seizure
19 (24.4)
On awakening
11 (14.1)
Status epilepticus
0
Non-seizure related AMS
59 (75.6)
Symptom duration
<10 min
10 (12.4)
EEG helped establish diagnosis
10-59 min
15 (18.5)
No
38 (48.7)
>59 min
56 (69.4)
Yes
40 (51.3)
Seizure history
EEG confirmed or ruled out pre-EEG diagnosis
No
61 (78.2)
Ruled out diagnosis
4 (9.8)
Yes
19 (21.8)
Confirmed diagnosis
37 (90.2)
Witnessed tonic clonic activity
EEG changed management in ED
No
54 (66.7)
No
75 (96.1)
Yes
27 (33.3)
Yes
3 (3.9)
Clinical signs of seizure
Change in management resulting from ED EEG
Ocular movement abnormalities
5 (6.2)
Started a therapy
1 (1.3)
Subtle motor activity in face/extremity
2 (2.5)
Stopped a therapy
2 (2.6)
Generalized convulsions
20 (24.7)
Ordered additional tests
0
Management did not change
75 (96.2)
Level of consciousness (LOC) at enrolment
Alert
23 (28.4)
Would have ordered EEG on this patient if routinely available in ED
Somnolent/confused
52 (64.2)
No
42 (53.9)
Stupor
5 (6.2)
Yes
36 (46.2)
Coma
1 (1.2)
CT head in ED
No
11 (13.6)
tistic using Eq. (2), where a is assumed to be .05. Cohen's kappa is
Yes
70 (86.4)
the most commonly used kappa-type statistic in epidemiological
Head CT findings
studies (Thompson and Walter, 1988).
Unremarkable
32 (45.7)
Acute disease process
6 (8.6)
Pa A Pe
Chronic disease process
32 (45.7)
K 1/4
1
1 A Pe
Anticonvulsant therapy prior to EEG
Ativan
4 (4.9)

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
s
ffi
pffiffiffi

Pa A 1 A Pa
Phosphenytoin
1 (1.2)
CI 1/4 K 1/2A11 A A 2 A errora


 A
2
n1 A Pe2
Keppra
1 (1.2)
Level of agreement for the Cohen's kappa values was considered
excellent (above 0.80), good (0.61-0.80), moderate (0.41-0.60) and
agreement and we assumed a probability of chance agreement,
fair to poor (below 0.40), as recommended by Landis and Koch
P(e), of .1. Confidence intervals were calculated for each kappa sta-
(1977).

---

W.C. Ziai et al. / Clinical Neurophysiology 123 (2012) 910-917
913
Table 6
nated as ``unable to interpret'' and a survey was not administered.
Univariate analyses of factors associated with positive utility of EEG in the ED and
The duration of time for various steps in the EEG acquisition pro-
whether EEG would be ordered if available in the ED.
cess (Table 3) includes all patients with completed EEGs. The
Factor
Positive utility of
EEG would be ordered
20 min EEG interpretation was normal in 39.0%, diffusely abnormal
ED EEG P-value
if available p-value
in 31.7%, focally abnormal in 17.1% and uninterpretable in 11.0%
Pre-EEG impression (seizure
0.15
0.03
(Table 4). No patient's EEG showed status epilepticus and one
vs. no seizure)
patient had a tonic clonic seizure during the recording.
Witnessed seizure activity
0.15
<0.001
Surveys were completed by ED physicians or physician assis-
Level of consciousness at
0.58
0.13
tants (PA) (Table 5). Of 78 completed surveys in patients with
presentation
Normal EEG vs. abnormal EEG
0.85
0.39
20 min recordings (including 6 uninterpretable EEGs), EEG helped
Focal abnormality on EEG
0.91
0.69
to determine a diagnosis in 51.3%, changed ED management in 3.9%
EEG interpretation
0.76
0.84
and would be ordered again if available in 46.2%. In 15/40 (37.5%)
Etiology of altered mental
0.01
0.08
cases in which the EEG was helpful for diagnosis, the ED practi-
status (pre-EEG)
Toxicologic/psychiatric/
0.001
0.93
tioner would not have ordered an EEG on that patient if routinely
metabolic cause of AMS
available; while in 11/38 (28.9%) cases in which the EEG was not
(pre-EEG)
helpful for diagnosis, the ED practitioner would have ordered an
Pharmacologic cause of AMS
0.03
0.21
EEG if available in the ED. The EEG was more likely to help confirm
(pre-EEG)
a diagnosis (90.2% of surveys) than to rule out a diagnosis (9.8%).
Traumatic brain injury history
0.17
0.02
(not present)
Changes in management included two patients in whom antiepi-
Survey taken by PA (vs. MD)
0.33
0.008
leptic therapy started in the ED was stopped and one patient
Pathology on head CT
0.37
0.87
who was started on an antiepileptic agent.
Time symptom onset before
0.16
0.29
In the univariate analysis (Table 6), positive utility of EEG (i.e.
arrival
Sudden onset of symptoms
0.03
0.10
EEG helped in making diagnosis) was significantly associated with
Symptom duration <1 h
0.44
0.006
pre-EEG etiology of AMS (p = 0.01), specifically toxicologic, psychi-
Patient age
0.21
0.17
atric and endocrine/metabolic causes (p = 0.001), as well as a phar-
macologic cause by itself (such as drug overdose or antiepileptic
drug toxicity) (p = 0.03), and a sudden onset of symptoms
(p = 0.03). Factors associated with whether ED EEG would be or-
Table 7
dered if available in the same patient were witnessed seizure activ-
Multivariate analysis of factors associated with positive utility of EEG in ED.
ity (p < 0.001), no antecedent head injury (p = 0.02), symptom
Factor
Odds
95% CI
p-
duration <1 h (p = 0.006), and survey respondent being a PA (vs.
ratio
value
MD) (p = 0.008). In the multivariate analysis (Table 7), factors which
Etiology of AMS per discharge diagnosis
1.35
0.94-1.92
0.10
were independently associated with positive EEG utility, adjusting
Toxic/psychiatric/metabolic cause of
34.14
5.48-
<0.001
for the discharge diagnosis of AMS, were a suspected toxic, psychi-
AMS
212.61
atric or metabolic cause of AMS (pre-EEG) (p < 0.001) and a sudden
Sudden symptom onset
6.87
1.69-27.96
0.007
symptom onset (p = 0.007). Factors which were independent pre-
dictors of whether an EEG would be ordered on a particular patient
if available in the ED were witnessed seizure activity (p = 0.01), ab-
sence of antecedent head injury (p = 0.001), and survey respondent
Table 8
being a physician assistant (vs. MD) (p = 0.02) (Table 8).
Multivariate analysis of factors associated with whether EEG would be ordered if
available in the ED.
Factor
Odds
95% CI
p-
3.2. EEG Quality
ratio
value
Witnessed seizure activity
6.59
1.46-
0.01
Fig. 1 includes sample EEG segments from each of the 6 inter-
29.63
pretable EEG categories. Fig. 2(a) and (b) compare the EEG quality
Antecedent traumatic brain injury
0.10
0.02-0.39
0.001
of the 20 min EEG with the abbreviated EEG (5 min EEG). Median
Symptom duration <1 h
2.61
0.75-9.04
0.13
scores were similar although interquartile range was smaller for
Survey respondent being a physician
3.71
1.27-
0.02
assistant
10.82
the 20 min EEG for both EEG waveform and artifact analysis. The
5 min EEG (vs. 20 min EEG) presented good agreement on wave-
form shape/amplitude (good or excellent vs. poor or satisfactory:
j = 0.78, 95% CI 0.68-0.88), presence of artifact (good or excellent
3. Results
vs. poor or satisfactory: j = 0.75, 95% CI 0.62-0.85) and good or
excellent agreement on all interpretation categories (all j values
3.1. Clinical characteristics and EEG utility
P0.70) (Tables 9 and 10). There were more uninterpretable and
normal EEGs reported in the 5 min EEG group and more focally
Over 1 year, 82 patients (mean age 58.1 2.0 years) underwent
abnormal EEGs and diffusely abnormal EEGs without epileptiform
ED EEG. Patient demographics and presenting features of AMS are
discharges reported in the 20 min EEG group (Fig. 3).
shown in Tables 1 and 2. Witnessed tonic clonic seizure activity
had occurred in 33%. History of seizures was present in 21/81
(22%). Nearly two thirds of patients were somnolent or confused
4. Discussion
at time of presentation to ED. The majority of patients (86%) under-
went head CT in the ED, of which 8.6% had acute findings. EEGs
The purpose of this study was to determine feasibility and util-
were obtained in 82 patients. One patient was too agitated to ob-
ity of performing EEGs in the ED using best current practices (full
tain an EEG (excluded from the study) and three cases were
montage 20 min EEG with interpretation by a trained professional).
aborted due to excessive technical artefact which was not thought
We determined that rapid availability of standard full-montage
to be patient related. These three EEGs were included, but desig-
EEG in the ED is feasible and helps to establish a diagnosis in about

---

914
W.C. Ziai et al. / Clinical Neurophysiology 123 (2012) 910-917
Fig. 1. Sample EEG segments from each of the six interpretable EEG categories used for interpretation.
half of patients with seizures or AMS at presentation, although it
icantly shortened if electronic transfer of EEG data had been used.
rarely changes management.
This was not done in the current study due to internet limitations
Feasibility was assisted by use of a commercially available head
in the emergency room. The application of the head cap and EEG
cap, applied by a trained EEG technician and immediate availability
acquisition could be taught to a core group of ED personnel to re-
of an EEG expert who reviewed the entire tracing at the bedside of
place the need for an EEG technician at the bedside. In addition, if
all, but one patient. While this is ideal, it is not practical in most
the EEGs were performed by the ED it would decrease the number
hospital settings and provides both challenges and opportunities
of unnecessary and unhelpful EEGs.
for improvement. The time interval from the completion of the
In this study the clinical characteristics which provided the
EEG to EEG expert arrival (approximately 20 min) could be signif-
most diagnostic utility of EEG in the ED (non-structural neurologic

---

W.C. Ziai et al. / Clinical Neurophysiology 123 (2012) 910-917
915
Fig. 2. (a) Comparison of full-montage 20 min EEG with full-montage 5 min EEG for waveform shape and amplitude (N = 81). (b) Comparison of full-montage 20 min EEG
with full-montage 5 min EEG for EEG artifact and noise (N = 81).
A relatively low percentage of EEGs were uninterpretable in the
Table 9
ED given the high risk setting for artifact and patient agitation. EEG
Agreement (adjusted for chance) between 20 min EEG and 5 min EEG interpretations
diagnostic categories with the best agreement between 5 and
for individual diagnostic categories.
20 min EEG were electrographic seizures, focal abnormalities (with
EEG diagnostic category
Kappa statistic (95%
Interpretation
or without epileptiform discharges) and diffusely abnormal EEG
confidence interval)
with epileptiform discharges. The categories of diffusely abnormal
Diffusely
0.86 (0.79-0.94)
Excellent
EEG without epileptiform discharges, normal EEG and uninterpret-
abnormal + epileptiform
able EEG only showed ``good'' agreement. Measures of EEG signal
discharges
Diffusely abnormal A no
0.70 (0.59-0.81)
Good
quality and artifact had good agreement. We propose several
epileptiform discharges
explanations. First, the 5 min EEG interpretation was based solely
Focal
0.89 (0.82-0.96)
Excellent
on the EEG display without knowledge of patient history, examina-
abnormalities + epileptiform
tion, or EEG-video monitoring. This does not reflect clinical reality,
discharges
Focal abnormalities
0.95 (0.89-0.99)
Excellent
where the EEG interpretation is typically made in association with
A no
epileptiform discharges
the above. In addition the 5 min period chosen between minutes
Electrographic seizures
0.99 (0.96 - 1.00)
Excellent
10 and 15 may have missed important EEG data since patients
Normal EEG
0.72 (0.61-0.82)
Good
who were sleeping were awoken at the end of the 20 min EEG to
Unable to interpret
0.73 (0.63-0.83)
Good
demonstrate the ``awake-EEG'' pattern. These factors would be ex-
pected to enhance the interpretation and may account for the dis-
crepancy in number of ``normal'' and ``diffusely abnormal'' EEGs
and possibly the higher frequency of ``uninterpretable'' assess-
Table 10
ments of the 5 min EEG.
Agreement (adjusted for chance) between 20 min EEG and 5 min EEG interpretations
A second issue is the classification scheme used for this study.
for individual categories of EEG quality indicators.
We became aware that there are varying degrees and types of arti-
EEG quality indicator
Kappa statistic (95% confidence interval)
fact that adversely affect to different degrees, the interpretation of
Category
Waveform shape/amplitude
Artifact/noise
the seven listed categories. For example, if just two electrodes were
1-3 vs. 4-5
0.78 (0.68-0.88)
0.75 (0.62-0.85)
non-functional for just half of the recording, but the rest were per-
Very poor
0.82 (0.81-0.96)
0.78 (0.68-0.88)
fect, we could not exclude focal status epilepticus. Alternatively If
Poor
0.82 (0.73-0.91)
0.70 (0.59-0.81)
98% of the EEG was uninterpretable across all channels, and the pa-
Satisfactory
0.71 (0.60-0.82)
0.64 (0.53-0.76)
tient was awake, then 2% could still determine if the patient had a
Good
0.60 (0.49-0.72)
0.66 (0.54-0.77)
normal background (i.e. no diffuse background slowing). Thus dif-
Excellent
0.71 (0.60-0.82)
0.82 (0.73-0.91)
ferent degrees, focalities, diffuseness and types of artifact variably
affected what could be excluded as an abnormality or confirmed as
causes of AMS (toxic, metabolic and psychiatric) and sudden onset
normal. The classification would therefore have to be modified to
of symptoms) were different from those which would prompt an
best reflect whether we are looking for background issues, focal is-
ED practitioner to order an EEG. This may in part reflect a learning
sues or epileptiform issues in isolation, or together. A possible
curve of the type of patient in whom EEG is useful. The patients for
improvement on the existing classification would require qualify-
whom ED practitioners would be most likely to order an EEG were
ing artifact in specific channels vs. all channels, and adjusting the
those who had witnessed seizures, symptom duration <1 hr and
EEG specific requirements for each diagnostic category. We believe
non-traumatic causes of AMS. This likely reflects both the concern
such improvements could reduce greatly the number of uninter-
to rule out ongoing seizures and the importance of distinguishing
pretable studies.
patients who had a seizure (showing post-ictal slowing on EEG)
A variety of strategies and devices have been tested for EEG
from those with psychiatric or other transient causes of AMS in
detection in the ED (Bautista et al., 2007; Naunheim et al., 2010)
whom the EEG should be normal. Practitioners were always in-
Bautista et al. used abbreviated (5 min) EEG in the ED for workup
formed that the EEG could not confirm whether a seizure had
of 25 patients with AMS and confirmed useful information was
definitively occurred.
obtained without interrupting the routine workup of such patients.

---

916
W.C. Ziai et al. / Clinical Neurophysiology 123 (2012) 910-917
Fig. 3. EEG interpretation: comparison of full-montage 20 min EEG with full-montage 5 min EEG (N = 82).
Although ED physician input was not obtained, they found 2/25
Financial Interests
(8%) patients with non-convulsive status epilepticus and found
that diffuse slowing on EEG was highly suggestive of diffuse
This work was supported by a Research Grant from the Epilepsy
encephalopathy or a non-neurologic event. Only two studies were
Foundation. This publication was also supported by a subcontract
uninterpretable. EEGs were performed by experienced EEG
from Infinite Biomedical Technologies, LLC (``IBT''). Dr. Ziai has
technicians.
received support from the Epilepsy Foundation, and has served in
The next stage is the development of an ED EEG algorithm
an editorial capacity for Current Opinion in Neurology. Dan Schl-
which could be developed as an aid to various EEG acquisition
attman also received support from the Epilepsy Foundation. San-
models including ED technician-driven EEG performance or the
tosh Venkatesha, Dan Schlattman, and Peter Kaplan have all also
rapidly developing field of automated electroencephalograms. As
received support from NIH Grant U44 NS057966-01. The remain-
more advanced detection systems are developed and brought into
ing authors have no conflicts of interest.
clinical use, a better definition of which clinical criteria are
important for maximizing the diagnostic yield of abbreviated
Acknowledgements
EEG systems will be necessary to yield clinically useful informa-
tion. While the absence of seizure activity may be useful in all
We thank the physicians, physician assistants and nurses of the
patients with AMS, the differentiation of normal from focal or dif-
Bayview Medical Center Emergency Department for their partici-
fuse abnormalities on EEG would be potentially more useful in
pation in this study and patience with the performance of EEGs
patients with psychiatric illness and possible drug overdose,
on patients under their care.
whereas the degree of abnormality of EEG background activity
would be more suited to assess severity of an underlying
encephalopathy.
References
The other major advancement not employed by this study is re-
American College of Emergency Physicians. Clinical policy for the initial approach to
mote monitoring, now available in many centers. Rapid on-line
patients
presenting
with
altered
mental
status.
Ann
Emerg
Med
detection of electrographic seizures and communication on EEG re-
1999;33:251-80.
sults to ED personnel would significantly reduce time delays and
American Clinical Neurophysiology Society. Minimum technical requirements for
performing clinical electroencephalography. J Clin Neurophysiol 2006;23:86-91.
could affect outcomes provided expert evaluation of EEG were
Bautista RE, Godwin S, Caro D. Incorporating abbreviated EEGs in the initial workup
immediately available. The cost-benefit ratio of introducing an
of patients who present to the emergency room with mental status changes of
ED EEG service with around the clock availability of a neurolo-
unknown etiology. J Clin Neurophysiol 2007;24(1):16-21.
Breen P, Murphy K, Browne G, Molloy F, Reid V, Doherty C, et al. Formative
gist/clinical neurophysiologist has not been assessed. Breen et al.
evaluation of a telemedicine model for delivering clinical neurophysiology
evaluated a telemedicine model for delivery of clinical neurophys-
services part I: utility, technical performance and service provider perspective.
iology services in a clinic setting and found a comparable unit cost
BMC Med Inform Decis Mak 2010;15(10):48-60.
Cohen J. A coefficient of agreement for Nominal scales. Educ Psychol Meas
per EEG compared to a traditional EEG service although there are
1960;20:37-46.
few similarities to an emergency department model (Breen et al.,
Erkinjuntti T, Wikstrom J, Palo J, Autio L. Dementia among medical inpatients:
2010). To determine benefit would require demonstration of ear-
evaluation of 2000 consecutive admissions. Arch Intern Med 1986;146:1923-6.
lier diagnosis of seizures, improved emergency room throughput
Hoosmand H, Maloney M. The role of the EEG in the emergency room. Clin
Electroencephalogr 1980;11:163-8.
and potentially shorter length of stay for patients with altered
Huff JS, Morris DL, Kothari RU, Gibbs MA. Emergency Medicine Seizure Study Group.
mental status. Alternatively, there may be other frequent emer-
Emergency department management of patients with seizures: a multicenter
gency room diagnoses for which early EEG takes on a more impor-
study. Acad Emerg Med 2001;8(6):622-8.
Johnson J, Sims R, Gottlieb G. Differential diagnosis of dementia, delirium and
tant role. The value of EEG monitoring in predicting prognosis after
depression: implication for drug therapy. Drugs Aging 1994;5:431-45.
cardiac arrest in the setting of therapeutic hypothermia is one such
Kanich W, Brady WJ, Huff JS, Perron AD, Holstege C, Lindbeck G, et al. Altered mental
application which has received recent interest (Rossetti et al.,
status: evaluation and etiology in the ED. Am J Emerg Med 2002;20(7):613-7.
Kaplan PW. Assessing the outcomes in patients with nonconvulsive status
2010).
epilepticus: nonconvulsive status epilepticus is underdiagnosed, potentially

---

W.C. Ziai et al. / Clinical Neurophysiology 123 (2012) 910-917
917
overtreated
and
confounded
by
comorbidity.
J
Clin
Neurophysiol
Rossetti AO, Urbano LA, Delodder F, Kaplan PW, Oddo M. Prognostic value of
1999;16:341-52.
continuous EEG monitoring during therapeutic hypothermia after cardiac
Landis JR, Koch GG. The measurement of observer agreement for categorical data.
arrest. Crit Care 2010;14(5):R173.
Biometrics 1977;33:159-74.
Thompson WD, Walter SD. A reappraisal of the kappa coefficient. J Clin Epidemiol
Meena AK, Prasad VS, Murthy JM. Neurological intensive care in India - disease
1988;41:949-58.
spectrum and outcome. Neurol India 2001;49(Suppl. 1):S1-7.
Wofford J, Loehr L, Schwartz E. Acute cognitive impairment in elderly ED patients.
Nandhagopal R. Generalised convulsive status epilepticus: an overview. Postgrad
Etiologies and outcomes. Am J Emerg Med 1996;14:649-53.
Med J 2006;82(973):723-32.
Wolfe RE, Brown DFM. Coma and depressed level of consciousness. In: Marx JA,
Naunheim RS, Treaster M, English J, Casner T. Automated electroencephalogram
Hockberger RS, Walls RM, editors. Rosen's emergency medicine: concepts and
identifies abnormalities in the ED. Am J Emerg Med Published online: 1-May-
clinical practice. 5th ed. St. Louis: Mosby, 2002. p. 137-44.
2010; doi:10.1016/j.ajem.2010.03.010.
Quigg M, Shneker B, Domer P. Current practice in administration and clinical
criteria of emergent EEG. J Clin Neurophysiol 2001;18:162-5.

---

Document Outline

• Emergent EEG in the emergency department in patients with altered mental states
  • Introduction
  • Methods
    • Design
    • EEG interpretation
    • Statistical analysis
  • Results
    • Clinical characteristics and EEG utility
    • EEG Quality
  • Discussion
  • Financial Interests
  • Acknowledgements
  • References

---