Vous êtes sur la page 1sur 8

Neurophysiologie Clinique/Clinical Neurophysiology (2015) 45, 97—104

Disponible en ligne sur

ScienceDirect
www.sciencedirect.com

REVIEW/MISE AU POINT

EEG guidelines in the diagnosis of brain


death
L’EEG dans le diagnostic de mort cérébrale

W. Szurhaj a,b,∗, M.-D. Lamblin a, A. Kaminska c, H. Sediri a

a
Service de neurophysiologie clinique, CHRU, 59037 Lille cedex, France
b
Faculté de médecine, Université Lille 2, Lille, France
c
Laboratoire de neurophysiologie clinique, Hôpital Necker—Enfants Malades, AP—HP, Paris, France

Received 5 November 2014; accepted 9 November 2014


Available online 14 January 2015

KEYWORDS Summary In France, for the determination and diagnostic validation of brain death the law
Electroencephalogra- requires either two EEG recordings separated by a 4-hour observation period, both showing
phy; electrocerebral inactivity; or cerebral angiography examination. Since EEG is available in most
EEG; hospitals and clinics, it is often used in this indication, at the patient’s bedside, especially
Brain death; in the context of organ donation. However, very precise methodology must be followed. The
Recommendations; last French guidelines date back to 1989, before the development of digital EEG recording. We
Technical aspects present the new guidelines from the Société de Neurophysiologie Clinique de Langue Française.
Electrocerebral inactivity may be confirmed when a 30-minute good quality EEG recording shows
complete electrocerebral silence, defined as no cerebral activity greater than 2 uV, having first
ruled out the possible influence of sedative drugs, metabolic disorders or hypothermia. In the
presence of sedative drugs, CT brain angiography will be the gold standard test for this diagnosis.
In the newborn, the utmost caution is indicated since electrocerebral inactivity can be observed
in the absence of cerebral death. In the infant, the criterion for the observation period to be
respected between both EEG recordings needs to be more clearly refined.
© 2015 Elsevier Masson SAS. All rights reserved.

Résumé En France, la loi exige que le diagnostic clinique de mort cérébrale soit con-
MOTS CLÉS firmé par deux EEG inactifs réalisés à 4 heures d’intervalle ou une angiographie cérébrale.
Électroencéphalo- La disponibilité de l’EEG dans la plupart des centres hospitaliers et sa facilité de réalisation
gramme ; au lit du patient en font toujours un examen très utilisé dans cette indication, notamment
EEG ; dans le cadre des prélèvements d’organes. Sa réalisation exige toutefois une méthodologie

∗ Corresponding author. Service de neurophysiologie clinique, CHRU, 59037 Lille cedex, France. Tel.: +33 06 76 76 32 07.
E-mail address: w.szurhaj@gmail.com (W. Szurhaj).

http://dx.doi.org/10.1016/j.neucli.2014.11.005
0987-7053/© 2015 Elsevier Masson SAS. All rights reserved.
98 W. Szurhaj et al.

précise. Les dernières recommandations françaises dataient de 1989, avant l’avènement de


Mort cérébrale ; l’EEG numérisé. Nous présentons ici les nouvelles recommandations de la Société de neuro-
Recommandations ; physiologie clinique de langue française. L’EEG sera considéré « inactif » lorsqu’un tracé de
Aspects techniques 30 minutes, de bonne qualité, montre un silence électrique, défini comme l’absence d’activité
cérébrale d’une amplitude supérieure à 2 ␮V, à la condition d’exclure l’influence possible de la
sédation, de troubles métaboliques et d’une hypothermie. En cas d’imprégnation par des sub-
stances sédatives, l’angioscanner sera préféré. En contexte néo-natal, une grande prudence
est nécessaire, un tracé inactif pouvant être observé en l’absence de mort encéphalique. Chez
le nourrisson, les critères d’intervalle entre les deux EEG devront encore être mieux définis.
© 2015 Elsevier Masson SAS. Tous droits réservés.

Abbreviations • EEG recording should include a minimum of eight scalp


electrodes (standard ones or ideally needle electrodes):
FP2, C4, O2, T4, FP1, C3, O1, T3. Cz is recommended;
• simultaneous ECG (in D1) and respiratory movement
CT computed tomography recording can help identify certain activities as artifacts.
ECG electrocardiogram It seems essential to be able to have these two recordings
EEG electroencephalogram during the entire EEG duration;
• impedances for each electrode must be between 100 and
10,000 . The recording will be performed at a sampling
Summary of guidelines rate of at least 256 Hz, and the high-pass filter should be
set at 0.53 Hz. Digital filters below 70 Hz should not be
It is mandatory for the physician interpreting EEG in the used (apart from the notch filter at 50 Hz), as this could
context of brain death to obtain the maximum amount of attenuate high-frequency brain waves;
data: • the EEG tracing should include at least 30 minutes of good
quality recording in order to be properly interpreted;
• before performing an EEG, essential clinical data must • auditory and bilateral somatosensory stimuli (touch and
be collected: patient’s last name, first name and date of pain) should be repeatedly performed, and be clearly
birth, circumstances and origin of the coma, especially indicated on the recording (for somatosensory stimuli,
date and time of the presumed onset of the coma, and indication of the stimulated point and excluding electro-
imaging results; static shocks);
• the clinical examination (neurologic assessment) must • the technician must note on the recording all events that
specify brain death criteria (deep coma; absence of spon- occur during the examination;
taneous respiration; absence of brainstem reflexes); • system calibration is usually automated in digital EEG;
• any clinical suspicion of pharmacological, biochemical however the integrity of the entire recording system
or toxic factors, as well as metabolic disturbance or should be tested by voluntarily creating artifacts (shaking
hypothermia should be mentioned in the evaluation the connecting leads of the electrodes for example);
report; • in order to validate the lack of electrical activity on the
• data regarding the EEG recording should include: date, EEG, we recommend maintaining the patient’s body tem-
time and location of the EEG recording and possibly scalp perature above 34 ◦ C (minimum recommended). In the
condition. Drugs administered in the preceding 24 hours or case of hypothermia, with a temperature below 34 ◦ C,
drugs screened for in blood and urine should be indicated, the patient’s body will need to be warmed up before
as well as body temperature and hemodynamic parame- proceeding with the EEG;
ters, especially mean arterial pressure, which should be • we recommend that the EEG be performed at least
maintained above 50 mmHg. 12 hours after cardiorespiratory arrest;
• it is necessary to repeat the EEG with a minimum
of 4 hours between the two examinations (minimum of
The execution parameters must be rigorously observed: 4 hours calculated from the beginning of each recording);
• since techniques based on the study of intracerebral blood
• the recording should be performed by a special- flow (cerebral angiography) are not influenced by seda-
ized technician, or physician who is board-certified in tive medications, we recommend using these techniques
electroencephalography, these being the only persons rather than EEG in this situation;
qualified to identify and reject artifacts, implement the • if it is impossible to use these alternative techniques, the
required polygraphy set-up, perform necessary activa- EEG interpretation must take into account the sedative
tion procedures and document all relevant data on the drugs previously administered, and eventually residual
recording; drug levels in the blood sample.
EEG guidelines in the diagnosis of brain death 99

The interpretation conditions should allow for the best both cerebral perfusion scintigraphy confirming irreversible
interpretation possible, without any possible objections: cerebral insult and EEG showing electrocerebral inactivity
are mandatory for the diagnosis of brain death. In Australia
• electrocerebral inactivity (term of reference rather than and in some Asian countries, clinical criteria based on the
using ‘‘flat-lined’’, ‘‘null’’, ‘‘isoelectric’’) is defined by absence of brainstem reflexes are applied.
the absence of cerebral activity above 2 ␮V amplitude, In summary, the major differences between criteria listed
even when auditory and nociceptive stimuli are per- in the different guidelines [27] are: the number of examining
formed in a bilateral manner; physicians required to confirm brain death, their seniority
• the interpretation of the EEG will be easier on a good and level of experience; additional examinations are some-
quality display screen with a minimum resolution of times mandatory and sometimes only recommended; the
1280 × 1024. The optimal screen size for proper reading duration of the observation period after declaration of brain
is 17 inches or more; death varies, the most usual being around 24 hours; the
• for an optimal interpretation of the recording, the reading apnea test is the object of controversy because of poten-
settings should range between 3 and 5 ␮v/mm (inclu- tially noxious consequences on internal organs [23].
sive) with a speed of 20 s per display screen/page (i.e. Some countries, in which organ transplantation is not per-
15 mm/s). A 30 mm/s speed (10 s per display screen) can formed (some African countries for example), do not have
be used to better appreciate the signal; brain death criteria.
• for paper traces, the printer should have a minimum res-
olution of 300 DPI; French laws
• the interpretation should be carried out using long dis-
tance montages, without filters;
In France, the last regulatory text still in application today
• the interpretation report can only be written by a physi-
was published in 1996. It governs the procedures and nec-
cian specialized in functional exploration of the central
essary conditions for the diagnosis of brain death and also
nervous system. EEG results should only be communicated
organ donation.
to physicians qualified to confirm a diagnosis of brain
Decree no 96-1041 of December 2, 1996 relative to the
death, and no one else.
diagnosis of death, prior to removing organs, tissues or cells
for therapeutic or scientific purposes and modifying the Pub-
In children, due to brain immaturity, specific recommen- lic Health Code.
dations should be taken into account: Section 3. —– Removing organs from a deceased patient.
Sub-section 1. —– Validation of brain death before remov-
• the clinical examination and precise brain death criteria ing organs for therapeutic or scientific purposes.
should be detailed on the EEG request and be repeated; Art. R. 671-7-1. —– If the person is in cardiac and respira-
• in premature babies and newborns under 7 days of life, tory arrest, brain death can only be validated if the three
because of brain immaturity, EEG is not a reliable indi- following clinical criteria are simultaneously observed:
cator of brain death and cerebral angiography should be
used; • lack of any evidence of consciousness and of spontaneous
• between the age of 7 days and 1 month, it is recommended motor activity;
to perform two EEGs separated by a 48-hour interval and • absence of all brainstem reflexes;
above all to take into account clinical criteria; • total absence of spontaneous respiration.
• beyond the age of 1 month and up to 1 year, the recom-
mended interval between the two EEGs is 24 hours;
Art. R. 671-7-2. —– If the person, whose death was clin-
• beyond the age of 1 year, the recommended delay
ically validated, is assisted via mechanical ventilation and
between two examinations is similar to that applied to
conserves hemodynamic functions, the absence of sponta-
adults (i.e. 4 hours);
neous breathing is verified using the hypercapnia test.
• under the age of 2 months, the interpreting physician
Furthermore, in addition to the three clinical criteria
must be experienced in performing EEG in newborns and
mentioned in the article R. 671-7-1, the following tests are
infants.
mandatory to validate the irreversible nature of the brain
insult:
Full-length guidelines
• either two electroencephalogram recordings showing
The utility of EEG in the diagnosis of brain death is regarded electrocerebral inactivity, respecting a 4-hour observa-
differently in different countries. In the United States, tion period between both tests, conducted with maximal
EEG is recommended but is not mandatory [16,25,26]. In amplification on a 30-minute recording; the result must be
Canada, EEG is no longer recommended for this indication. immediately recorded by the physician who interpreted
In Europe, an EEG laboratory test to validate the diagnosis of the recording;
brain death is not recommended in all countries. In several • or cerebral arteriography showing lack of blood flow to the
countries, the apnea test is performed, to look for hyper- brain; the radiologist who interpreted the imaging exam
capnia based on an increase of PCO2 > 20 mmHg compared must immediately record the result.
to normal. The duration of the observation period between
the initial event and the EEG varies from 0 to 48 hours, espe- Art. R. 671-7-3.-III. —– When brain death has been val-
cially when anoxia is the cause of brain death. In Japan, idated for a person undergoing mechanical ventilation
100 W. Szurhaj et al.

retaining hemodynamic functions, the brain death report Artifacts caused by resuscitation conditions (mains
indicates the results of the concordant clinical observations power, monitor screens, ballistocardiogram, etc.) can usu-
of two physicians adhering to the conditions mentioned in ally be easily eliminated. However, it is sometimes not
article L. 671-10. Furthermore, it mentions the result of possible to eliminate all artifacts, and the interpreting
examinations as defined in the 1st or 2nd addendum of arti- physician may therefore be unable to confirm the electro-
cle R. 671-7-2, as well as the date and time of the report. cerebral inactivity of the EEG recording. This explains in part
The report is signed by the two physicians mentioned above. the lack of sensitivity reported for EEG in certain studies.
The EEG in this last text appears as one of the mandatory In some rare circumstances, such as open traumatic brain
examinations, along with cerebral arteriography (or more injury, EEG cannot be performed due to obvious technical
recently cerebral CT-angiography), necessary to confirm the difficulties.
irreversible and definite nature of the brain destruction.
Specific conditions of duration, amplification and observa-
tion period between the examinations are clearly defined, EEG specificity
but refer to traditional EEG paper recordings.
The study conducted in the 1960s by the American EEG
Advantages and limits of EEG in the early Society’s Ad Hoc Committee on EEG criteria for the deter-
diagnosis of brain death mination of cerebral death revealed that, of 2650 cases of
coma with presumably ‘‘isoelectric’’ EEGs, only three whose
records satisfied the committee’s criteria showed any recov-
The main advantage of EEG is its non-invasive, reproducible
ery of cerebral function. These three patients had received
nature, and it is easily available in most hospitals.
massive overdoses of nervous system depressants (barbitu-
In a study conducted in France in 1997, C. Fischer [9]
rates in 2 cases and meprobamate in one case). This study
reported that in the diagnosis of brain death, physicians
did not include children [1].
used one or more investigations in addition to clinical exam-
However, electrocerebral inactivity requires verification
ination and that in 100% of cases the investigation was an
of the absence of certain conditions that could affect the
EEG, followed in 67% of cases by brain arteriography. Two
degree of brain insult:
EEG recordings seemed to be the most reliable method to
validate brain death (as reported by 60% of physicians).
• EEG is highly sensitive to drugs, especially barbiturates,
EEG sensitivity which can lead to depression of electrogenesis or even
a silent trace. This situation requires postponing the
EEG until sedative drugs have been cleared from the
Very few recent data are available.
patient’s body. The drugs concerned are barbiturates,
The study by Paolin et al. in 1995 [18] showed a rela-
benzodiazepines, anesthetics, especially propofol but
tively low sensitivity of EEG in a population of 15 adults
also trichloroethylene, methoqualone, meprobamate and
meeting the clinical criteria of brain death (absence of
baclofen; the latter have a very long half-life and require
brainstem reflexes and absence of spontaneous respira-
a long period of time before being completely eliminated
tion): EEG recording showed electrocerebral inactivity in
from the patient’s system [10,14,15,19,21];
only 8 patients and showed residual low-voltage activity in
• hypothermia is known to alter CNS function. It can slow
the other 7 patients (53% sensitivity). However, this study
down or depress electrogenesis but mainly in major
was performed in a small number of subjects using paper
hypothermia, below 28 ◦ C [7]. Coselli et al. [6] attempted
EEG devices; furthermore, the results seem discordant with
to correlate the apparition of an ‘‘isoelectric’’ recor-
other studies on this topic. In the study by Grigg [11] on
ding with peripheral body temperature, in 56 patients
56 patients, meeting the clinical criteria of brain death,
operated under hypothermia. They observed a wide
the sensitivity was better (80.6%). Three aspects were
temperature variability for the onset of ‘‘isoelectric’’
observed: low-voltage theta or beta activity in 9 patients
recording, with values ranging from 12.8 ◦ C to 28.6 ◦ C
(observed up to 72 hours after the clinical diagnosis), a
(rectal temperature). Furthermore, hypothermia can
sleep-coma appearance in two patients (up to 168 hours
potentially alter the metabolism and clearance of certain
after the clinical diagnosis), and alpha coma in one patient
administered medications.
(3 hours after having met the clinical criteria). None of these
patients recovered. This study also involved only paper EEG
recordings. Unfortunately, to our knowledge, there is no study that
Progression to digital EEG, sometimes coupled with video has clarified the impact of hypothermia on electrogenesis in
recording, has facilitated the identification of artifacts, the specific context of brain death.
and thus probably greatly improved the sensitivity of EEG. Most experts agree that hypothermia > 30 ◦ C cannot be
This was demonstrated in a recent German study [24], the cause of electrocerebral inactivity, yet in the absence of
which compared somatosensory evoked potentials, EEG, validated data, most guidelines require a body temperature
transcranial Doppler, radionuclide cerebral perfusion study above 35 ◦ C.
and cerebral angiography in 71 patients meeting the criteria Some metabolic disorders can have an impact on
of brain death. EEG was able to validate brain death (elec- brain function: acute liver failure, severe hypoglycemia or
trocerebral inactivity) in 67 cases out of 71 (94% of cases), hyponatremia. Since these conditions are reversible, it is
similar to cerebral CT-angiography results and much better important to correct them or to repeat the EEG before con-
than the other techniques. firming electrocerebral inactivity.
EEG guidelines in the diagnosis of brain death 101

Apart from these specific situations, EEG specificity is of the filter effect of the pen and galvanometer, which
very good. Only a few cases of electrocerebral inactivity only detects a signal in the order of 1 ␮V whereas digital
have been reported in patients who were not clinically dead, EEG detects amplitudes of 0.15 ␮V (i.e. 10 times more sen-
these having been observed before the widespread use of sitive). This point was confirmed by a second study that
digital EEG. consisted of generating a sinusoidal signal with modulat-
ing frequency and amplitude, highlighting that there was a
Reproducibility of the EEG major attenuation of the recorded signal for fast frequencies
in conventional paper-based EEG, whereas no attenuation
was observed for digital EEG, even for frequencies as high
Buchner and Schuchardt [4] evaluated intra- and inter-
as 100 Hz. In the light of these results, the authors pro-
examiner reproducibility: they submitted 28 EEG recordings
posed technical guidelines, in addition to the criteria listed
of patients with clinically validated brain death, to 8 neurol-
in French regulatory texts and those produced during the
ogists, experienced in reading EEG; furthermore physicians
consensus conference for acquisition and reading of elec-
had to interpret the same recording twice without being
trocerebral activity on digital EEG recordings.
informed of this.
These guidelines were published by Sediri et al. in 2007
Eighteen of the 28 recordings were interpreted in a
[20]. They complemented the 1989 guidelines from the
concordant manner, reflecting electrocerebral inactivity,
SNCLF [5].
and 3 out of the 28 were interpreted as retaining some corti-
cal electrical activity. Seven out of 28 recordings (25%) were
the subject of discordant decisions. Required information before conducting the
Furthermore, each physician interpreted in a different EEG
manner at least 1 recording between the 2 interpretations
they carried out on the same EEG. Before proceeding with the EEG recording all relevant data
This study was based on paper EEG recordings. To our must be collected: for example, patient’s first name, last
knowledge there are no recent data on the reproducibility name and date of birth, circumstances and origin of the
of digital EEG in the context of brain death. coma, especially presumed time and date of coma onset
EEG is thus still a highly relevant investigation in the and imaging results.
diagnosis of brain death. However, in France, because of The clinical examination report must detail the clinical
the need to perform 2 tests respecting a 4-hour observa- brain death criteria (deep coma, absence of spontaneous
tion period, the influence of drugs on the recording and the respiration, absence of cranial nerve reflexes).
fact that EEG is not available outside of routine working Any clinical suspicion of pharmacological, biochemical or
hours in most hospitals, this test has become less used in toxicological factors as well as metabolic factor or hypother-
the diagnosis of brain death, cerebral CT-angiography being mia should be mentioned.
the examination of choice when available. Data regarding the EEG recording must include:

EEG recording conditions for the diagnosis of • date, time and location of the EEG recording and eventu-
brain death ally status of the scalp;
• drugs administered in the past 24 hours or systematically
screened for in blood and urine samples;
In France, guidelines pertaining to EEG recording conditions
• body temperature and hemodynamic vital signs, espe-
in the confirmation of brain death were defined during the
cially mean arterial pressure, which should be maintained
conference of the Société de Neurophysiologie Clinique in
above 50 mmHg.
1988, published in 1989 [5]. These recommendations were
defined at a time when EEG recording was still acquired on
paper-based devices. Technical requirements
Since the 1990s, technological advances have allowed the
emergence of digital EEG, which has progressively replaced, The recording must be conducted by a specialized technician
in routine practice, the older devices in most hospitals. Since or electrophysiology physician, these being the only per-
then it has been necessary to redefine the criteria of brain sons qualified to identify and discard artifacts, implement
death for digital EEG devices. appropriate means of recording, perform the necessary
A compared analysis of paper and digital EEG recordings stimulation procedures and note all relevant data on the
was carried out simultaneously in 15 patients diagnosed recording.
with brain death [8]. This study highlighted increased util- The EEG recording must include at least 8 electrodes
ity for the digital EEG recording vs. paper EEG recording (standard electrodes or, ideally, needle electrodes): FP2,
for one particular patient. The presence of fast pseudo- C4, O2, T4, FP1, C3, O1, T3, as well as a ground electrode;
micro-rhythms, with an amplitude around 2 ␮V presented adding a median electrode is recommended. The resistance
a challenge for interpretation, and in this case digital EEG of each electrode must be between 100 and 10,000  [1].
seemed more sensitive than conventional paper-based EEG. Simultaneous ECG (in D1) and respiratory monitoring can
This activity was extracerebral ‘‘noise’’ and was generated help to identify some activity as artifacts. It is essential to
by the electrical components of digital devices, interference have these simultaneous recordings during the entire EEG
coming from the environment and monitoring equipment recording. Recording electrocerebral activity from one sin-
(e.g. electrodes and amplifier). Paper-based EEG devices did gle channel, as is sometimes done in the ICU, is insufficient
not have the capacity to transcribe these activities because for the diagnosis of electrocerebral inactivity.
102 W. Szurhaj et al.

Recording should be conducted at a sampling frequency The interpretation of the EEG will be easier with a good
of at least 256 Hz, a resolution of at least 12 bits and with a quality screen with a minimal resolution of 1280 × 1024. The
high-pass filter set at 0.53 Hz. No digital filter below 70 Hz ideal screen size for reading EEG recordings is 17 inches or
(aside from the 50 Hz filter) must be used since this could more [20]. The vertical resolution will depend on the number
attenuate the fast cortical activity frequencies. of channels displayed (at least 10 including 8 for the EEG,
In the case of persistent muscular artifacts, which could 1 ECG and 1 respiratory). Flat screens have an advantage in
interfere with the interpretation of the EEG trace, it may be terms of clarity. The printer must have a minimum resolution
necessary to inject a neuromuscular blocking agent (curare) of 300 DPI.
under the supervision of an anesthesiologist or physician The interpretation can only be performed by a physi-
familiar with the drug. cian specialized in functional exploration of the central
In case of persistent artifacts, with potential doubt as nervous system. The result must be given to those physi-
to their cerebral origin, it might be useful to add an addi- cians authorized to deliver the diagnosis of brain death, and
tional extracerebral channel, for example using 2 electrodes to them only. The validation of the irreversible nature of
positioned on the patient’s hand. CNS alterations is based on the observation of persistent
The EEG trace should include at least 30 minutes of good electrocerebral inactivity on the EEG recording.
quality recording, in order to be interpreted without ambi- The EEG should yield 30 minutes of stable, clean and
guity. easily interpretable inactive recording. Electrocerebral
Visual, auditory, somatosensory (tactile and painful inactivity is defined as the absence of EEG activity over
stimuli) stimulation procedures must be repeatedly per- 2 ␮V, even during auditory and nociceptive stimulations
formed and be clearly indicated on the recording (for procedures performed on both sides, without any possible
somatosensory stimuli, indicating the point stimulated and objection.
discarding electrostatic shocks). The technician must note Specific conditions of EEG interpretation in the diagnosis
on the EEG tracing all stimulation procedures performed of brain death are discussed below.
during the EEG recording.
The system calibration is usually fully automated in digi-
Hypothermia
tal EEG, but the integrity of the entire recording system must
be tested by voluntarily creating artifacts (for example by
In France, the Société Française d’Anesthésie et de Réani-
touching each electrode).
mation, the Société de Réanimation de Langue Française,
Artifacts encountered in the intensive care unit must be
and Biomedicine Agency recommend a body temperature
identified and if possible eliminated. The ballistocardiogram
equal to or above 35 ◦ C before establishing the diagnosis of
created by the ECG and diffusing to all EEG derivations, is
brain death. Some countries require significant absence of
often the most difficult to eliminate. It is possible to remove
hypothermia, with a temperature above 32 ◦ C, but they do
this artifact during recording if the patient’s head is kept
not use EEG. Moderate hypothermia with a body tempera-
turned to one side; if present, the interpreting physician
ture above 30 ◦ C cannot yield an inactive EEG recording, but
will take this into account when reading the EEG, making
it could increase the abnormalities on the recording.
sure that the artifact produced is perfectly synchronous to
We recommend a body temperature equal to or above
the QRS complex noted on the ECG channel. Other arti-
34 ◦ C in order to be able to confirm electrocerebral inac-
facts are caused by electrical parasites related to the 50 Hz
tivity on the EEG. In the case of hypothermia below 34 ◦ C,
power line noise and other critical care monitoring devices
the patient will have to be warmed up before performing
in the patient’s room. This type of artifact can be removed
the EEG recording. Further studies will be relevant to assess
by using a reliable ground power outlet, unplugging air-
the impact of hypothermia between 30 and 34 ◦ C on brain
fluidized beds and choosing a different power outlet from
electrogenesis in the diagnosis of brain death.
the other monitoring devices used for patients’ critical care.
The reverberation artifact can be eliminated by not placing
the EEG device in front of other monitors present in the Observation period
room.
French laws do not mention a specific observation period
after the initial event before validating the diagnosis of brain
Interpretation death, in contrast to other countries, mainly European ones
[13]. However, the EEG can be misleading in showing false-
The EEG interpretation will be based on long distance bipo- positive electrocerebral inactivity in the hours immediately
lar montages with inter-electrode distances of 10 cm. We following a cardiorespiratory arrest, before the reappear-
recommend a montage including frontal, central, temporal, ance of cerebral activity. For this reason, we recommend
occipital, right and left electrodes, for example: Fp2-C4, that EEG be performed respecting a 12-hour observation
C4-O2, Fp1-C3, C3-O1, Fp2-T4, T4-O2, Fp1-T3, T3-O1. period after the cardiorespiratory arrest.
Reading settings should range between 3 and 5 ␮v/mm In France, the law requires performing repeat EEG test
(inclusive) with a speed of 20 s per display screen/page (i.e. with a 4-hour observation period between both tests, which
15 mm/s). A 30 mm/s speed (10 s per display screen) can be is sometimes difficult in a patient presenting with hemo-
used to better appreciate the signal. dynamic instability. The 4-hour observation period between
High-frequency filters should not be set below 70 Hz, and both EEG tests is indeed one of the main reasons why physi-
low-frequency filters should not be set above 0.5 Hz. The cians nowadays tend to prefer cerebral CT-angiography in
notch filter set at 50 Hz can be used. this context.
EEG guidelines in the diagnosis of brain death 103

Very few data from the literature are available regarding The American Academy of Pediatrics (AAP) recently
changes observed between two consecutive EEG recordings (2011) published its guidelines [17]. In the United States, the
in this context. All scientific societies, including the Ameri- diagnosis of brain death is made solely on clinical criteria
can Academy of Neurology (AAN), do not make it mandatory and the apnea test. ‘‘Ancillary studies (electroencephalo-
to perform two separate EEG tests. It seems quite unlikely gram and radionuclide cerebral blood flow) are not required
that an EEG with electrocerebral inactivity, performed in to establish brain death and are not a substitute for the neu-
the absence of sedation, hypothermia, or major metabolic rologic examination. However ancillary studies may be used
disorders, would be significantly modified after a 4-hour to assist the clinician in making the diagnosis of brain death
observation period; however the test needs to be performed (i) when components of the examination or apnea testing
at a distance from the initial condition that led to the clin- cannot be completed safely due to the underlying medical
ical diagnosis of brain death. The need to repeat the EEG condition of the patient; (ii) if there is uncertainty about the
test, when the latter has been performed at an appropri- results of the neurologic examination; (iii) if a medication
ate distance from the initial event, could be challenged. We effect may be present; or (iv) to reduce the inter-
recommend that further studies be conducted to evaluate examination observation period’’. Experts all agree that the
the changes in EEG recording between two repeat tests. The younger the child the longer the inter-examination obser-
goal would be to avoid the 4-hour inter-examination obser- vation period should be between the clinical examinations
vation period, which can sometimes limit the use of EEG in leading to the diagnosis of brain death: AAP recommends an
certain situations, and can delay organ harvesting. observation period of 24 hours for term newborns (37 weeks
While further studies are awaited to clarify these issues, gestational age) to 30 days of age, and 12 hours for infants
we recommend repeating the EEG test while respecting and children (> 30 days to 18 years). Performing an ancillary
an inter-examination observation period of at least 4 hours exam (for example EEG) can reduce this observation period.
between both tests (measured from the beginning time of Thus, the Société Française d’Anesthésie Réanimation pub-
each EEG). lished its guidelines in 2005 [3]:

• ‘‘in children, the diagnosis of brain death is based on dif-


Medication
ferent criteria than in adults, after having excluded, just
like in adults, confounding factors [3];
The question of drug plasma levels for sedative drugs admin-
• in preterm newborns and term newborns under 7 days of
istered to the patient is an important and difficult issue,
life, because of cerebral immaturity, EEG is not a reliable
and is not mentioned in the French regulatory text of 1996.
indicator of brain death, and physicians should usually rely
Some drugs, such as barbiturates, can have a long clear-
on cerebral angiography;
ance level according to the drug’s half-life, and as such
• between the age of 7 days and 2 months, it is recom-
it is not always possible to wait for drug plasma levels to
mended to perform two clinical examinations and 2 EEGs
return to zero. The American Academy of Neurology (2005)
with an inter-examination observation period of 48 hours,
stipulates: ‘‘If intoxicants such as barbiturates, benzodi-
except in cases of anoxic brain injury;
azepines, or opioids are present, levels need not be zero, but
• between the age of 2 months and 1 year, it is recom-
should be in a range that would not normally be expected
mended to perform two clinical examinations and 2 EEGs
to interfere significantly with consciousness’’. Other scien-
with an inter-examination observation period of 24 hours,
tific societies, and especially the Biomedicine Agency in its
except in cases of anoxic brain injury;
latest report on removing donated organs recommends sim-
• beyond the age of 1 year, criteria are similar to those
ply that drug plasma levels be below therapeutic range.
applied to adults.’’
Techniques based on cerebral blood flow, not being influ-
enced by sedative drugs, are recommended rather than
EEG in this context. Furthermore, the study of evoked Some studies [2] suggest that there is no evidence to
potentials (auditory and somatosensory) is an electro- justify using different observation periods in adults and chil-
physiological technique less sensitive to sedative drugs dren above the age of 1 month. As a matter of fact, there
and hypothermia and could help determine the absence are no reported cases of children who recovered neurologi-
of cortical and sub-cortical activity, even though it has cal functions after having met all the criteria of brain death
no medical-legal value in France today [12]. applied to adults [17].
When it is impossible to use these alternative ancillary A review of the literature based on 12 studies (published
exams, it is preferable to wait for drug plasma levels results in the guidelines of the American Academy of Pediatrics
to come back before interpreting the EEG and wait for these in 2011) [17], on a cohort of 485 children suspected of
levels to be validated as not interfering in a significant man- brain death, is reported in these guidelines. It evaluated
ner in cerebral electrogenesis. the sensitivity of EEG at 76% in children; this sensitivity
increased to 89% with a repeat EEG test. Two false-positive
tests were reported each time children were being given
Specificities for the diagnosis of brain death in barbiturates. The sensitivity of EEG was as good as cerebral
children CT-angiography in children over the age of 1 month, but it
was lower in the group of children under the age of 1 month
In France, there is no specific regulatory text for the diag- (40% vs. 63%).
nosis of brain death in children. However experts agree that Based on these data, we have established the following
adult criteria cannot be applied to young children [22]. guidelines:
104 W. Szurhaj et al.

• the clinical examination and specific brain death criteria [6] Coselli JS, Crawford ES, Beall Jr AC, Mizrahi EM, Hess KR,
must be noted on the EEG request and must be repeated; Patel VM. Determination of brain temperatures for safe circu-
• in preterm newborns and term newborns < 7 days of life, latory arrest during cardiovascular operation. Ann Thorac Surg
because of cerebral immaturity, EEG is not a reliable 1988;45:638—42.
indicator of brain death, and it is recommended to use [7] Danzl DF, Pozos RS. Accidental hypothermia. N Engl J Med
1994;331:1756—60.
cerebral CT-angiography;
[8] Derambure P, Caillez S, Bourriez JL, JDG. Utilisation de
• between the age of 7 days and 1 month, it is recommended l’EEG numérisé dans le diagnostic de la mort encéphalique.
to perform two EEG tests, respecting a 48-hour inter- L’évaluation neurophysiologique des comas, de la mort
examination observation period; encéphalique et des états végétatifs. Solal E, editor: Guérit
• between 1 month and 1 year: the recommended inter- JM; 2001. p. 271—87.
examination observation period is 24 hours; [9] Fischer C. [The use of EEG in the diagnosis of brain death in
• above the age of 1 year: the recommended inter- France]. Neurophysiol Clin 1997;27:373—82.
examination observation period is identical to that of [10] Garcia-Larrea L, Fischer C, Artru F. [Effect of anesthetics on
adults. sensory evoked potentials]. Neurophysiol Clin 1993;23:141—62
[Effet des anesthésiques sur les potentiels évoqués sensoriels].
[11] Grigg MM, Kelly MA, Celesia GG, Ghobrial MW, Ross ER. Elec-
However these recommendations on age and observation troencephalographic activity after brain death. Arch Neurol
periods are based on empirical knowledge, and we recom- 1987;44:948—54.
mend conducting further studies to provide evidence-based [12] Guérit JM. Apport pronostique des potentiels évoqués en unité
data, in order to better refine ages and observation periods. de soins intensifs. Ann Fr Anesth Reanim 2004;23:99—101.
In the specific context of infants, the interpreting physi- [13] Haupt WF, Rudolf J. European brain death codes: a comparison
cian must be experienced in neonatal EEG. of national guidelines. J Neurol 1999;246:432—7.
[14] Meinitzer A, Kalcher K, Gartner G, Halwachs-Baumann G, Marz
W, Stettin M. Drugs and brain death diagnostics: determination
Conclusion of drugs capable of inducing EEG zero line. Clin Chem Lab Med
2008;46:1732—8.
EEG is a test used to validate brain death as listed in regu- [15] Mellerio F. EEG changes during acute intoxication with
trichlorethylene. Electroencephalogr Clin Neurophysiol 1970;
latory texts, adding evidence to the clinical criteria of brain
29:101.
death, but in no circumstance can it be a substitute to these [16] Morenski JD, Oro JJ, Tobias JD, Singh A. Determination of
clinical criteria. Two EEG recordings, respecting a 4-hour death by neurological criteria. J Intensive Care Med 2003;18:
inter-examination observation period, are necessary. 211—21.
In this situation, electrocerebral inactivity is consid- [17] Nakagawa TA, Ashwal S, Mathur M, Mysore M, Society of Crit-
ered when EEG during a 30-minute, good quality recording ical Care Medicine SoCC, Section on Neurology of American
evidences electrical inactivity defined as the absence of Academy of P, et al. Clinical report-Guidelines for the deter-
cerebral activity above 2 ␮V, excluding the possible influ- mination of brain death in infants and children: an update
ence of sedation, metabolic disorders or hypothermia. of the 1987 task force recommendations. Pediatrics 2011;128:
In case of impregnation by sedative drugs, cerebral CT- e720—40.
[18] Paolin A, Manuali A, Di Paola F, Boccaletto F, Caputo P, Zanata
angiography will be the examination of choice. Criteria for
R, et al. Reliability in diagnosis of brain death. Intensive Care
the diagnosis of brain death in newborns and infants still Med 1995;21:657—62.
need to be better clarified. [19] Powner DJ. Drug-associated isoelectric EEGs. A hazard in brain-
death certification. JAMA 1976;236:1123.
[20] Sediri H, Bourriez JL, Derambure P. Place de l’EG dans le
Disclosure of interest
diagnostic de mort cérébrale. Rev Neurol (Paris) 2007;163:
248—53.
The authors declare that they have no conflicts of interest [21] Sullivan R, Hodgman MJ, Kao L, Tormoehlen LM. Baclofen over-
concerning this article. dose mimicking brain death. Clin Toxicol 2012;50:141—4.
[22] Vecchierini-Blineau MF, Moussalli-Salefranque F. [Diagnosis of
brain death in the newborn and the child]. Neurophysiol
References Clin 1992;22:179—90 [Diagnostic de la mort cérébrale chez le
nouveau-né et l’enfant].
[1] American Clinical Neurophysiology Society. Guideline 3: min- [23] Vivien B, Haralambo MS, Riou B. Barotraumatisme lors du test
imum technical standards for EEG recording in suspected d’apnée chez des patients en état de mort encéphalique. Ann
cerebral death. Am J Electroneurodiagnostic Technol 2006; Fr Anesth Reanim 2001;20:370—3.
46:211—9. [24] Welschehold S, Boor S, Reuland K, Thomke F, Kerz T, Reuland A,
[2] Ashwal S. Brain death in the newborn. Current perspectives. et al. Technical aids in the diagnosis of brain death: a compari-
Clin Perinatol 1997;24:859—82. son of SEP, AEP, EEG, TCD and CT angiography. Deutsch Arztebl
[3] Boulard G, Guiot P, Pottecher T, Tenaillon A. Prise en charge Int 2012;109:624—30.
des sujets en état de mort encéphalique dans l’optique [25] Wijdicks EF. Determining brain death in adults. Neurology
d’un prélèvement d’organes. Ann Fr Anesth Reanim 2005;24: 1995;45:1003—11.
836—43. [26] Wijdicks EF. Practice parameters for determining brain death
[4] Buchner H, Schuchardt V. Reliability of electroencephalogram in adults (summary statement). The Quality Standards Sub-
in the diagnosis of brain death. Eur Neurol 1990;30:138—41. committee of the American Academy of Neurology. Neurology
[5] Clinique Bdlsdn. Recommandations quant aux conditions de 1995;45:1012—4.
réalisation d’un enregistrement EEG exigible pour le constat [27] Wijdicks EF. Brain death worldwide: accepted fact but no global
d’une mort cérébrale. Neurophysiol Clin 1989;19:339—41. consensus in diagnostic criteria. Neurology 2002;58:20—5.