Keywords Continuous electroencephalogram monitor - depth of anaesthesia monitor - electroencephalogram
- monitoring - quantitative electroencephalogram monitor
INTRODUCTION
Electroencephalography (EEG) is believed to represent a composite of the postsynaptic
potentials of the pyramidal cells of the cerebral cortex. For years, it has been an
invaluable diagnostic tool for neurologists in the diagnosis and management of patients
with epilepsy.
Since the first observations of the effects of anaesthetic agents on EEG in the year
1937 by Gibbs et al .,[1 ] EEG has provided a huge amount of information to the anaesthesiologist, either for
routine clinical practice as depth of anaesthesia monitors and for detection of sub-clinical
seizures; or for understanding the complex mechanisms of anaesthesia-induced alteration
of consciousness.
In initial years of its development, the routine clinical applicability of EEG within
the anaesthesia field was hindered by the complexity of the raw EEG signal and requirement
of a neurophysiologist for its interpretation. However, with technological advancement,
several EEG-derived dimensionless indices have been developed that correlate with
the depth of the hypnotic component of anaesthesia and are easy to interpret. Similarly,
with the development of quantitative EEG (QEEG) tools, the routine use of continuous
EEG (cEEG) is ever expanding in the Intensive Care Units (ICUs).
In this review, we have described various commonly used EEG-based monitors and their
clinical applicability in the field of anaesthesia and critical care.
Raw electroencephalogram
The raw EEG is a complex random-looking waveform and has many constituent oscillations
of varying amplitude and frequency. The general synchrony of the EEG appears to be
related to a pacemaker-like influence from the thalamus or brainstem.[2 ]
The four frequently described frequency bands of the normal EEG include: Delta (0–3
Hz), theta (4–7 Hz), alpha (8–12 Hz), and beta (13–30 Hz) waves [Table 1 ]. A band of higher frequency is also described and is referred to as gamma waves
(>30 Hz).
Table 1
EEG waves and their interpretation
Wave
Frequency (Hz)
Description
Explanation
*RAS = Reticular activating system, EEG = Electroencephalogram, GABAergic = Gamma-aminobutyric
acid-ergic
Gamma (γ) waves
>30
Higher frequency waves, markedly depressed by anaesthetics
Are thought to play a role in perception and processing of sensory information
Beta (β) waves
13-30
High frequency, low amplitude, awake state, usually seen in the prefrontal regions,
often seen to increase during the initial stages of sedation, amnesia and anxiolysis
Result of sensory stimuli influencing the RAS of the brainstem which releases inhibition
of the nucleus reticularis and desynchronizes the thalamic pacemakers resulting in
higher frequency content in the thalamo-cortical pathways
Alpha (α) waves
9-12
Medium frequency, higher amplitude, awake but eyes closed, more prominent over the
vertex
Related to the rhythmic influence of the thalamic pacemaker cells when the inhibitory
influence on the thalamus by the nucleus reticularis is reduced by meditation or anaesthesia
Theta (θ) waves
4-8
Low frequency, normally seen in sleep and under general anaesthesia, normal in children,
abnormal in the awake adult
Arise from inhibition of the normal thalamic pacemaker cells by GABAergic action of
the reticular nucleus
Delta (δ) waves
0-4
Very low frequency, normal in infants and in stages 3 and 4 of sleep, seen in depressed
functions (coma, deep anaesthesia, hypoxia, ischaemia, infarction, poor metabolism)
Delta wave with sleep spindle pattern occurs when the thalamo-cortical system is hyperpolarized,
and connectivity between the environment and the brain is lost
In general, majority of anaesthetic drugs alter the EEG by producing an initial excitatory
stage characterised by desynchroniation and increased relative power of faster frequencies.[3 ] The fast beta (β) activity is most prominent in the frontal regions and moves posterior
as the anaesthetic drug effect increases. Concomitantly, a prominent area of EEG synchronisation
in the alpha range (8–13 Hz) develops over the more posterior regions and moves to
the frontal regions (referred to as the ‘anterior topographic shift’ or ‘frontal dominance’).
Increasing anaesthetic drug doses further causes progressive slowing until the EEG
achieves burst suppression (where periods of EEG activity are interspersed with periods
of a flat EEG) and finally, electrical silence or a flat EEG. Next the variability
(or entropy) in the EEG decreases as the EEG becomes synchronised with the thalamic
pacemaker. Finally, the electromyography (EMG) of the frontal muscles becomes quiescent
as the anaesthesia deepens.
Based on these observations the EEG may be capable of reflecting a gradual change
in anaesthetic effect by using a combination of EEG amplitude, frequency, variability,
topography and frontal EMG.
Processed electroencephalogram
Different methods of EEG processing have been developed to extract the EEG patterns
that reflect the anaesthetic effect.[4 ] The general principle governing construction of EEG-derived dimensionless indices
is based on the extraction of parameters from the EEG whose value is known to be statistically
correlated with anaesthetic agent concentration and/or clinically assessed depth of
hypnosis. With the availability of small digital computers, EEG signal processing
could be accomplished by converting the analog EEG signal to a digital signal (digital
domain) and using the computer to decompose the EEG into numeric components (QEEG).
The raw EEG is acquired by electrodes applied to the forehead overlying the frontal
cortex, an area of the central nervous system (CNS) known to be involved in perception
and memory/amnesia formation. The EEG signal is then filtered and amplified, digitised
and sent to the device for mathematical processing. The raw EEG is usually divided
into time segments (such as 2- to 10-s periods/epoch) and processed as segments. The
processing creates a normalised index, which is generally a number varying between
0 (no activity) and 100 (awake) or a letter indicating anaesthetic stage.
One of the main processing tools is Fourier transformation. This transformation is
based on the Fourier theorem, which essentially states that any signal can be represented
by a series of sine waves of different amplitudes, frequencies and phases (the phase
information describes if component waves are in synchrony with each other). Thus,
a complex EEG tracing can be deconstructed and expressed as the spectrum of its component
frequencies.[5 ]
Several other parameters can be extracted from the dissection of the EEG signal into
its frequency components, and from the estimation of the relative contribution of
each component (according to the frequencies and their amplitudes) to the global signal,
namely the power spectrum of the EEG.[6 ] Examples of such parameters are the peak frequency (the one with the highest power),
the median frequency (50% of power is achieved by lower frequencies, and 50% by higher
ones), and the spectral edge frequency (SEF) (a specified percentage of the power
is achieved by lower frequencies, generally 90% or 95%).
The SEF has been suggested to be an easily recognised parameter that follows depth
of anaesthesia.[7 ] Trending the SEF provides easily recognised changes in SEF and quicker recognition
of activation or depression of the EEG. Hence, SEF can be used as an ‘early warning’
indicator of impending changes in depth of anaesthesia.
The density spectral array (DSA) provides for recognition of dominant frequencies
over time, displayed as increased density of data points during each epoch (2–4 s)
of raw EEG analysis. The display updates over time so that one may follow changes
in dominant frequencies throughout the course of an anaesthetic, surgical, or pharmacologic
manipulation.
Bispectral analysis (also referred to as the bispectrum) is a mathematical technique
that incorporates the phase relationships between waves by determining the phase coupling
between different frequencies. This method compares different frequencies of EEG activity
looking for synchrony. Deepening of anaesthesia results in increasing synchrony because
of the decreasing numbers of independent pacemakers within the brain.
The burst suppression ratio (BSR) is defined as the percent time of flat line EEG
in a given interval (approximately 60 s). The burst suppression pattern is easily
recognised on the EEG and does not require complex proprietary algorithms to be measured.
It is associated with a lower metabolic rate and may be therapeutic, or unintentionally
induced as a result of high anaesthetic drug administration. Burst suppression, therefore
may be a reflection of harm, therapeutic or unintentional depending on the setting.
A number of processed EEG (pEEG)-based monitors have been introduced to clinical practice.
Currently available monitors with dimensionless indices include:
Bispectral index
Bispectral index (BIS) has been the first and most studied EEG-based monitor till
date. Its calculation algorithm involves power spectrum, relative activity in the
β frequency range, burst suppression activity, synchronised fast slow activity, and
bispectrum.[8 ]
Entropy
Entropy measures the predictability (or the sinusoidal nature) of 2 wide EEG frequency
bands: 0.8–30 Hz for state entropy (SE) and 0.8–47 Hz for response entropy (RE).[9 ]
SE ranges between 0 and 91 which is mainly based on cortical EEG activity, and is
supposed to reflect the depth of the hypnotic component of anaesthesia. RE, in addition,
takes account of facial EMG activity, which increases in case of non-counterbalanced
noxious stimulation. RE ranges between 0 and 100 and its gradient with SE has been
proposed to reflect the balance between nociception and anti-nociception during anaesthesia.[10 ]
Patient state index
The patient state index (PSI) is derived from 4 EEG channels and takes account of
the frontal localisation of power spectrum changes during anaesthesia.[11 ] The PSI algorithm incorporates relative activity in specific frequency bands, inter-hemispheric
coherence information, as well as anteroposterior frequency and phase relationships.
The output of the algorithm is a dimensionless number between 0 and 100.
Cerebral state index
The cerebral state monitor is a battery operated handheld device that analyzes a single
EEG channel and presents a cerebral state index (CSI) scaled from 0 to 100. CSI calculation
involves α ratio (natural log of the ratio of EEG energy in the 30–42.5 Hz band to
the EEG energy in the 6–12 Hz band), the β ratio (natural log of the ratio of the
energy in the 30–42.5 Hz band to the energy in the 11–21 Hz band), the difference
between them, and the amount of burst suppression (BS%) in each 30-s period.[12 ]
Narcotrend
The Narcotrend monitor analyzes stages and sub-stages of anaesthesia, and is based
on similar developmental process as the BIS (although with a distinct algorithm).[13 ] Six different visually recognisable EEG patterns were initially identified as corresponding
to different sleep stages. They were further divided into a total of 15 different
patterns evidenced during anaesthesia. Each of these stages was then characterised
by a set of EEG parameters, including spectrum, entropy, and autoregression. The output
is a number between 0 and 100, as well as a letter corresponding to the anaesthetic
stage.
Wavelet-based anaesthetic value for central nervous system monitoring
A wavelet analysis, which captures both time and frequency domain information on the
EEG signal in the form of wavelet coefficients, is used to calculate the wavelet-based
anaesthetic value for CNS index, ranging from 0 to 100.[14 ] The reaction time of that index is shorter than the reaction time of BIS.
SNAP index
The SNAP II is a battery operated EEG monitor that calculates a ‘SNAP index’ from
a single channel of EEG. The index calculation is based on a spectral analysis of
EEG activity in the 0–20 Hz and 80–420 Hz frequency ranges, and a burst suppression
algorithm.[15 ] The SNAP index ranges in value from 100 to 0.
Auditory-evoked potential index
The auditory-evoked potential index necessitates the administration of 7-Hz auditory
clicks to the patient through earphones. Middle latency auditory evoked potentials
are extracted from the raw EEG using a moving averaging window. The index is then
calculated based on the amplitude and latency of those evoked potentials, knowing
that amplitude decreases and latency increases with hypnotic depth.[16 ]
INCORPORATION OF PROCESSED ELECTROENCEPHALOGRAM-BASED MONITORS INTO CLINICAL PRACTICE
INCORPORATION OF PROCESSED ELECTROENCEPHALOGRAM-BASED MONITORS INTO CLINICAL PRACTICE
Prevention of intraoperative awareness
The incidence of awareness is believed to be between 0.1% and 0.2% in all surgical
cases and is as high as 1–10% in subspecialties such as obstetric, cardiac and trauma
surgery.[17 ] In 2012, guidelines were published by the National Institute for Health and Care
Excellence in the UK recommending the use of EEG-based brain monitoring, especially
in ‘vulnerable’ patients.[18 ]
[19 ]
The pEEG monitors currently available are able to discriminate awake from deeply anesthetised
patients showing burst suppression with an effectiveness of over 90%.[20 ] However, most anaesthesiologists currently do not use these monitors either through
choice or unavailability of the devices. All the more, the available large six randomised
controlled trials,[21 ]
[22 ]
[23 ]
[24 ]
[25 ]
[26 ] undertaken until date to demonstrate any potential benefit of using these monitors
for prevention of unexpected intraoperative awareness, show conflicting results.
According to Avidan and Mashour, an in-depth analysis of these studies leads to the
conclusion that an EEG-driven administration of intravenous hypnotic anaesthetic agents
(with pharmacological paralysis) helps prevent unexpected intraoperative awareness
with explicit recall, at least in patients at higher risk of experiencing such an
event. However, when volatile anaesthetic agents are used, monitoring their end-tidal
concentration, which reflects concentrations attained in the brain, and setting low
concentration alarms is at least as efficient at achieving the same goal. When general
anaesthesia is administered without neuromuscular blocking agents, movement might
be the best indicator of awareness.[27 ]
Monitors that combine spontaneous pEEG indices with metrics based on evoked potentials
are currently under development.[28 ]
[29 ] In such a monitor, analysis of spontaneous EEG signals may complement analysis of
evoked EEG signals in predicting whether a patient will react to a surgical stimulus.
Guiding anaesthetic management
EEG-based technologies may play an increasingly important role in the maintenance
of general anaesthesia within the therapeutic window and tailored to the intraoperative
needs of each patient. pEEG monitoring provides a combined appraisal of the CNS response
to the surgical stimulation,[30 ] cardiovascular function, respiratory function and drug action.[31 ] However, to be used as effective anaesthetic titration aids, EEG monitors should
allow individual drug titration, avoid episodes of over dosage, and hasten recovery
when the procedure has come to its end with less side effects. Clinical studies attempting
to clarify the role of pEEG monitors as anaesthetic titration aids, have given conflicting
results. A meta-analysis revealed a mean reduction in anaesthetic drug consumption
of 19% for BIS monitoring.[32 ] Ellerkmann et al . cited a linear correlation between the mean titrated BIS value and the hypnotic
drug saving potential.[33 ] In contrary, pEEG monitoring did not result in significantly reduced doses of volatile
anaesthetics in both the B-aware and B-unaware trials.[22 ]
[23 ]
In future, the refinement of ‘closed-loop’ systems in which anaesthetic delivery is
controlled based on feedback from BIS values may aid in maintaining appropriate depth
of anaesthesia.[34 ]
[35 ]
As monitors of ‘functional cerebral perfusion’
Both PSI and BIS seem to follow functional cerebral perfusion during periods of hypotension
and resuscitation, as well as the level of hypnosis and/or depth of anaesthesia.[36 ]
[37 ]
[38 ] Clinical use of the BIS or PSI should include observations of the DSA and SEF to
follow changes in underlying EEG frequencies contributing to these descriptors. Such
changes need to be correlated with ongoing surgical, anaesthetic, pharmacologic, and/or
physiologic manipulations.
During carotid endarterectomy (CEA) under general anaesthesia, the BIS falls and follows
changes in evoked potentials.[39 ] It is, however, unclear whether this is reliable as BIS has been reported a poor
predictor of ischaemia in awake CEA[40 ]
[41 ] or outcome under general anaesthesia CEA.[42 ]
During periods of controlled hypotension or in trauma surgery, the appearance of elevated
BSR values may indicate inadequate cerebral perfusion and alert the anaesthesiologist
to an impending ischaemic event.[43 ] Similarly, an increase in BSR values, even at ‘normal mean arterial pressure (MAP)’
in elderly and chronic hypertensive patients may indicate inadequate cerebral perfusion.
During ‘pharmacological brain protection’
The BSR is used clinically to quantify the duration of flat line EEG during attempts
to provide pharmacologic brain protection with propofol, barbiturates, or etomidate.[44 ]
[45 ]
[46 ] Significant theoretical reductions in CMRO2 may be obtained when burst suppression is twice the duration of active EEG as approximately
60% of CMRO2 is used in producing EEG. This 2:1 (suppression: burst) ratio may reduce CMRO2 by 40% (2/3 of 60), provided the suppression duration is at least 10 s.[47 ]
Assessing the balance between noxious stimulation and anti-nociception
BIS and EMG variability, or an index combining both parameters (the composite variability
index, ranging between 0 and 100), has been proposed for assessing the adequacy of
anti-nociception during anaesthesia.[48 ] BIS variability can be estimated using the standard deviation of BIS values recorded
during the previous 3 min. EMG variability corresponds to the standard deviation of
EMG power over the same period of time. The algorithm for its calculation has been
designed according to the ability of sub-parameters to predict the occurrence of a
somatic event (or patient movement) in response to noxious stimulation, in absence
of muscle relaxation. Another EEG-derived parameter proposed to assess anti-nociception
is the aforementioned RE-SE gradient of Spectral Entropy.[49 ]
Optimising peri-operative outcomes
Effect on early recovery
Amidst conflicting study results, the role of pEEG guidance in improving early recovery
outcomes remains unclear. In a meta-analysis, Liu, reported a modest reduction in
anaesthetic consumption, risk of post-operative nausea and vomiting (PONV), and recovery
room time in patients undergoing ambulatory anaesthesia with use of BIS monitoring.[32 ] Later, another Cochrane meta-analysis also found that BIS monitoring significantly
reduced propofol and volatile anaesthetic consumption, early recovery times, and length
of post-anaesthesia care unit (PACU) stay.[50 ]
However, in a large, randomised clinical trial Pavlin et al . reported that although BIS monitoring was associated with a slight decrease in sevoflurane
administration, it did not lead to faster emergence or a shorter stay in the PACU.[51 ] Similarly, analyses from the BAG-RECALL and B-unaware trial populations and the
Michigan Awareness Control Study demonstrated no difference in anaesthetic administration,
time to discharge from the post-operative recovery area, or incidence of PONV with
the use of BIS guidance compared with controls.[26 ]
[52 ]
Effect on post-operative mortality
An association has been found between prolonged time with low EEG indices during anaesthesia
and poor long-term outcomes, but not with absolute amount of anaesthetic agent received.[53 ]
[54 ]
[55 ]
[56 ] Whether these findings are related directly to low BIS scores or reflect an intrinsic
sensitivity of the brain to the effect of anaesthetic agents in vulnerable patients
is not yet clear.
In 2012, Sessler et al . reported findings from a large observational study of the possible influence of
low MAP, low BIS and low minimal alveolar concentration (MAC) on 30-day post-operative
mortality.[57 ] Authors found that low BIS (<45) in isolation was actually associated with a decreased
risk of 30-day mortality whereas, when low BIS occurred concurrently with the ‘triple
low’ (low MAC, low MAP, or both low MAC and low MAP), there was an increased adjusted
risk of 30-day mortality. Currently, large clinical trials are underway to determine
whether triple low or deep anaesthesia (e.g., low BIS) do indeed contribute to negative
post-operative outcomes.
Understanding mechanisms of general anaesthesia
EEG studies, either alone or in combination with other techniques such as transcranial
magnetic stimulation, have contributed to the considerable progress that has recently
been made in the understanding of the mechanisms of anaesthesia-induced alteration
of consciousness.[58 ]
[59 ] The advantage of EEG over other technique, is that it allows exploring functional
effective connectivity, or causal influence, between brain regions rather than statistical
dependencies in changes of indirect signs of regional brain activity.[60 ]
As a diagnostic tool in patients with epilepsy
The EEG can be the most helpful test to determine a diagnosis of epilepsy. It can
also distinguish focal and generalised neurophysiologic correlates of epilepsy.[61 ] Furthermore, when paired with video monitoring, EEG can not only define epileptic
and non-epileptic events but also aid in localisation of seizures in patients with
epilepsy.
Invasive intraoperative monitoring by electrocorticography [Figure 1 ], after placing intracranial depth wires or subdural grid electrodes, may be required
in patients with refractory focal epilepsy with the common goal of better defining
the cortical areas to be resected (ictal onset localisation) and/or to minimise the
risks of post-operative unacceptable deficits by cortical mapping of functional cortex
with electrical stimulation (usually speech, motor, sensory, and visual systems).[62 ]
[63 ]
Figure 1: Intraoperative monitoring by electrocorticography
CONTINUOUS ELECTROENCEPHALOGRAM MONITORING IN THE INTENSIVE CARE UNIT
CONTINUOUS ELECTROENCEPHALOGRAM MONITORING IN THE INTENSIVE CARE UNIT
In the neurocritical care setting, the clinical examination is often clouded by coma,
sedatives, toxic encephalopathy, or disorders of primary neurological modalities.
Continuous EEG [Figure 2 ] provides information about brain electrical activity, even when brain function is
depressed, and thus has a rapidly expanding role in the context of monitoring, early
diagnoses of sub-clinical seizures and non-convulsive status epilepticus, treatment
and prognostication among critically ill-patients with impaired consciousness [Table 2 ]
[64 ]
[65 ]
Figure 2: Continuous electroencephalography monitoring
Table 2
Role of cEEG monitoring in the ICU
Medical illness
Prevalence of seizures (%)
Importance of cEEG monitoring for
Seizure detection
Monitoring/treatment
Prognosis
Tentative grading of importance of cEEG = + = Minor importance, ++ = Moderate importance,
+++ = Strong importance, cEEG = Continuous electroencephalogram, ICU = Intensive Care
Unit, SAH = Subarachnoid haemorrhage
Intracerebral haemorrhage
10-30
++
+
Uncertain
SAH
4-16
++
++ for vasospasm
++
Traumatic brain injury
22-33
++
++
++
Ischemic stroke
5
+
Uncertain
+
Post-cardiorespiratory arrest/hypoxic-ischemic encephalopathy
10-33
+++
+++
+++
In medical ICU patients with severe sepsis
11
+
Uncertain
Uncertain
Focal EEG slowing may indicate ischaemia, whereas global slowing is suggestive of
encephalopathy. For prognostication purpose, reduced percentage of alpha variability
and absent EEG reactivity may indicate severe neuronal injury and poor prognosis.
The neurointensive care section of the European Society of Intensive Care Medicine
recommends cEEG monitoring in patients with generalised convulsive status epilepticus
and to rule out non-convulsive seizures (NCS) in brain-injured patients and in comatose
ICU patients without primary brain injury who have unexplained and persistent altered
consciousness. The panel also suggests EEG to detect ischaemia in comatose patients
with subarachnoid haemorrhage and to improve prognostication of coma after cardiac
arrest.[66 ]
In addition, EEG monitoring in ICU is also useful to help assess the depth of sedation,
monitor barbiturate coma for refractory intracranial hypertension[67 ] and is mandatory in several countries for the diagnosis of brain death.[68 ] It could also be useful for the early diagnosis of acute anoxic-ischemic brain injury
or intracranial hypertension cerebral damage; this type of indication is still under
evaluation.[64 ]
[69 ]
TECHNICAL CONSIDERATIONS
Periodic discharges (PDs) including (pseudo) periodic lateralised epileptiform discharges
(PLED), bilateral independent pseudoperiodic lateralised discharges, generalised periodic
epileptiform discharges, and triphasic waves (TWs) are patterns often encountered
in ICU EEG.[70 ] Technical choices such as types of electrodes, montages, as well as parameters of
quantified analysis are based on clinical indications, the patient’s state of consciousness
and the expected total duration of the EEG monitoring. Classic disc electrodes (Ag-AgCl)
positioned on the scalp with water-based adhesive paste, or held by a mesh net or
bandage, are usually recommended. Its disadvantage lies in the lengthy set-up time
and the generation of artifacts if a computed tomography scan or magnetic resonance
imaging becomes necessary.
Needle electrodes, faster to install and triggering less artifacts, are more invasive
and must be restricted to specific situations (e.g. diagnosis of brain death, coma,
limited access to the scalp).[71 ] A minimum of 8-electrode montage (electrode positions defined according to simple
skull landmarks while respecting the 10/20 international system) is recommended, that
can help detect up to 93% of epileptic seizures.[72 ] In addition, concomitant EEG video recording helps evaluate electro-clinical correlations
of seizures and identify the origin of artifacts.
To facilitate the interpretation of cEEG, digital trend curves based on various signal
characteristics can be displayed synchronously with the EEG. Most common used are
amplitude-integrated EEG (a-EEG) and spectral array ([DSA] or compressed spectral
array [CSA]). Visualised on one sole display graph, these trend curves can facilitate
the identification of very slow changes in EEG background activity and their variation
(alertness cycles, changes linked to treatment administrations) as well as seizure
patterns and their quantification. Spectral analysis from each hemisphere may also
be compared to provide an asymmetry spectrogram to assess the difference in amplitude
between cerebral hemispheres for each frequency.
In the a-EEG, EEG is compressed over time and adjusted in amplitude on a semi-logarithmic
scale. Specific a-EEG devices are commonly used in the neonatal ICU.[73 ]
INDICATIONS FOR CONTINUOUS ELECTROENCEPHALOGRAM MONITORING IN THE INTENSIVE CARE UNITS
INDICATIONS FOR CONTINUOUS ELECTROENCEPHALOGRAM MONITORING IN THE INTENSIVE CARE UNITS
In neurocritical care patients
cEEG monitoring is useful in identifying sub-clinical seizures and NCS and guiding
anti-epileptic therapy.
The diagnosis of NCS is essential in the ICU, especially in unexplained disorders
of consciousness, since their presence has been correlated with a poorer prognosis
and a delay in diagnosis appears to increase patient mortality.[74 ]
In the absence of clinical symptoms and EEG monitoring, most of the NCS would go unrecognized
as differentiating ongoing seizure activity from postictal or medication-induced encephalopathy
can be quite challenging in comatose ICU patients. In patients with acute neurological
diseases, NCS were detected in 21% of patients.[75 ] Among these patients, EEG diagnosis of NCS was preceded by clinical seizures only
in 25% of cases, whereas subtle clinical findings, such as oral or ocular muscular
movements and/or gaze deviation, were found in 50%.[75 ] cEEG is recommended over intermittent EEG because of the intermittent nature of
occult seizures.[66 ] Using cEEG recording, 56% of seizures are detected in the 1st h, and 88% in the first 24 h.[76 ]
cEEG also helps guide status epilepticus treatment, especially in case of refractory
status epilepticus treated by deep sedation by anaesthetic agents.[77 ]
In critically ill-patients with primarynon-neurological illness
EEG monitoring is required for early diagnosis of NCS and improves coma prognostication
after critical illness and cardiac arrest. NCS and PD are common in patients with
severe sepsis and altered mental status, and are associated with poor functional outcome.[69 ] However, a lack of EEG reactivity was found to be associated with higher 1-year
mortality than the presence of PD or NCS (mean survival time 3.3 [95% CI: 1.8–4.9]
vs. 7.5 [6.4–8.7] months; P = 0.002) in patients with severe sepsis.[65 ]
After cardiac arrest, post-anoxic encephalopathy needs to be differentiated from status
epilepticus. EEG is required as most seizures after cardiac arrest are non-convulsive
and to differentiate myoclonic status epilepticus from peripheral or subcortical myoclonus.[78 ] In addition, hypothermia and sedation used during cooling alter motor response and
decrease the prognostic accuracy of neurological examination.
EEG can reveal different type of abnormalities: Burst-suppression, spikes, spikes-and-waves,
polyspikes-and-waves, diffuse polyspikes or TWs, often periodic or pseudoperiodic.
The presence of discontinuous and burst-suppression pattern and of non-reactive EEG
background were found to be strongly correlated with a poor prognosis, while a continuous
reactive background was associated with good recovery.[79 ]
[80 ] Other automatic approaches using EEG-based evaluation of auditory functions, the
so-called mismatch negativity – an automatic fronto-central EEG component occurring
at 100–150 ms after the onset of a sound deviation has also been tested for prognostication.
One study showed that an intact auditory processing was present even in comatose patients
with extended brain damage after cardiac arrest, whereas a deterioration of auditory
discrimination over time was highly predictive of poor outcome.[81 ]
To detect cerebral ischaemia
EEG has become widely adopted for the detection of SAH-related vasospasm. Preliminary
studies have demonstrated that cEEG may become a potential useful tool in the monitoring
of cerebral vasospasm, allowing rapid diagnosis (24–48 h prior to other diagnostic
tools) and therapy before irreversible brain damage occurs.[82 ]
[83 ]
[84 ] In these studies, quantitative cEEG variables such as changes in total power, relative
alpha variability, the alpha/delta ratio and novel variable measuring alpha power
and variability, termed composite alpha index have been evaluated as sensitive predictors
of the development of vasospasm. An advantage of cEEG over other monitors of cerebral
vasospasm (e.g., transcranial Doppler ultrasonography and cerebral angiography) is
that it can be monitored continuously allowing an early therapy.
In acute ischaemic stroke, QEEG provides a good assessment of stroke size and severity.[85 ]
[86 ] In patients with haemorrhagic stroke, PLEDs are often observed in cortical haemorrhages
and are correlated with a poorer prognosis.[87 ]
To monitor barbiturate coma for refractory intracranial hypertension
An additional role for EEG in guiding therapy is in the management of refractory intracranial
hypertension. Although still controversial, due to the effects of hypotension and
infection,[88 ] barbiturate sedation should be performed in conjunction with EEG to minimise the
amount of medication by treating only until burst suppression has been obtained. Despite
being a common practice in the ICU, the duration of such suppression has yet not been
determined.
Limitations of electroencephalogram-based monitors
The accurate interpretation of EEG-derived indices is limited by several confounding
factors, which may lead to erroneous conclusions. They include inter-individual variability
in terms of baseline EEG characteristics,[89 ] erroneous site of EEG recording, artifacts from surrounding electrical devices (electrocautery,
pacemakers, warming devices, EMG),[90 ]
[91 ]
[92 ] and specific clinical conditions such as hypothermia, hypoglycaemia, dementia, cortical
atrophy, advanced age, seizures and cerebral ischaemia.[40 ]
[93 ]
[94 ]
[95 ]
[96 ] Several studies have reported that BIS values are less reliable in infants and small
children.[97 ]
[98 ] Ketamine and nitrous oxide tend to increase the BIS score and increase depth of
anaesthesia.[99 ]
[100 ] BIS values have been shown to decrease in patients receiving neuromuscular blockers,
both while fully conscious[101 ] or while anesthetised.[102 ]
Aquisition and interpretation of cEEG in ICU is limited by a number of factors including
wounds or bandages (that limit EEG electrode placement), sweating, muscle activity,
and movements commonly seen in delirious or agitated patients. Electrical interference
may occur from mechanical ventilators, machines for renal replacement therapy, neuromonitoring
apparatus, pumps, and electronic beds.
CONCLUSION
EEG provides a large amount of information to the anaesthesiologist, both for routine
clinical practice and for the understanding of the mechanisms of anaesthesia. Incorporating
pEEG monitors into routine anaesthetic practice can improve anaesthetic management
and patient outcomes by optimising the depth of anaesthesia. However, all interpretations
of monitored parameters require inspection and consideration of potential artifacts
prior to any clinical decision-making. In future, advances in EEG analysis may provide
the means to develop specific monitors of cortical connectivity, and particularly
of fronto-parietal feedback connectivity, which is thought to be a sign of consciousness.
In ICU settings, cEEG can convey a tremendous amount of physiologic information and
has the power to detect both ictal and inter-ictal EEG patterns. Future clinical trials
using cEEG and other real-time, continuous monitors (non-invasive or invasive) will
be useful and crucial to understand primary brain dysfunction as well as to reduce
the risk of secondary brain injury in susceptible critically ill-patients.
Financial support and sponsorship
Nil.
Conflicts of interest
There are no conflicts of interest.
