The basis of acceleromyography is that force is the product of mass and acceleration.³ During the assessment of neuromuscular blockade, mass is relatively unchanged so acceleration is proportional to force. An accelerometer comprises a small mass suspended on a strain gauge within a ‘box’ attached to the accelerating object to be studied, the acceleration then being derived from the force exerted on the strain gauge.

A number of accelerometric devices are available, such as the TOF-Watch (Organon Teknika, Fig. 17.2).

Figure 17.2 The TOF-Watch monitor (Organon Teknika). Stimulating current from 0–60 mA is set by the user. Accelerometric assessment of neuromuscular blockade or nerve localization may be performed. For the former, the device is calibrated after induction and before administration of muscle relaxant. (1) Stimulating electrodes applied over ulnar nerve; (2) accelerometric transducer taped to thumb; (3) display screen indicating train-of-four ratio: when less than four responses are detected, the number is indicated instead.

Piezoelectric methods (‘Kinemyography’)

When compressed or distorted, piezoelectric materials produce a charge proportional to the degree of alteration in shape. The M-NMT module for the S/5 anaesthesia monitor (GE Healthcare) uses a piezoelectric sensor incorporated into a clip placed on the patient’s thumb and index finger (Fig. 17.3). Thumb movement on stimulation is converted to an electrical signal and displayed graphically as a proportion of the maximal response. Monitoring, therefore, needs to be started before neuromuscular blockade but after anaesthetic induction. The module initially determines the current required for a supramaximal stimulus and sets a reference response level. By using different patient interface attachments the module can also be used for electromyographic recording or for nerve localization.

Figure 17.3 The mechanosensor for the Datex–Ohmeda M-NMT neuromuscular transmission module. (1) Stimulating electrodes applied over ulnar nerve; (2) piezoelectric mechanosensor.

Other methods

Mechanomyography, formerly the ‘gold standard’ for the assessment of neuromuscular blockade, has long disappeared from the clinical arena, as has electromyography. They were described in previous editions of this textbook. The general view is that, the above less-cumbersome methods are adequate for clinical purposes.⁴

Comparative studies suggest that the methods above do not give interchangeable results in assessing response to a train of four (TOF) stimulus.⁵ TOF measurements indicate residual blockade more consistently if compared with baseline readings, which should be obtained before paralysis.^6,⁷ Some authors recommend the use of ‘preload’ on the thumb to improve the accuracy of the measurement.⁴

Irrespective of the technology for assessing muscular response to nerve stimulation, numerous differing patterns of stimulation (e.g. train of four ratio, double burst stimulation, post-tetanic count, etc.) can be used in order to garner greater information regarding the extent and character of neuromuscular blockade. The details of these are in the remit of textbooks of physiology and of general anaesthesia.

Nerve stimulators for regional anaesthesia

Despite the growing use of ultrasound (see Chapter 13), there remains a need for peripheral nerve stimulators designed specifically for localization of nerves during regional anaesthesia (see Fig. 13.7). A positive surface electrode is applied to the skin and the negative electrode cable is plugged into a unipolar stimulating needle.

User-adjustable constant current is necessary and the ability to adjust stimulus frequency and duration is desirable. Usually an initial current of 1–2 mA (compare with currents needed for transcuteaneous stimulation of ∼60 mA) with a frequency of 2 Hz is appropriate as the needle is advanced toward the nerve. Lower frequencies increase the risk of going past the nerve, but may be desirable to reduce discomfort secondary to movement in injured patients.

Once contractions of the relevant muscle are seen, the current can be reduced. The stimulator-delivered charge required to generate a nerve impulse is proportional to the square of the distance from needle tip to nerve (by Coulomb’s law). Advantage can be taken of this phenomenon during neuraxial blockade in children, where a nerve stimulator attached to an insulated epidural needle may help determine whether the tip is epidural or intrathecal.⁸ For peripheral nerve blocks, traditional teaching is that injection of local anaesthetic is likely to be effective at the point where a response is still seen at a stimulating current of ≤0.3 mA.⁹ Recent work, however, suggests a similar probability of successful nerve block in paediatric patients whether the stimulation threshold is below or above 0.5 mA.¹⁰

Stimulus duration is adjustable on many devices. This should generally be set at less than 300 µs to avoid repetitive stimulation, though up to 1.0 ms may be required in neuropathic nerves, even with high currents.¹¹ Because chronaxie (the minimum stimulus duration required to elicit a response) is longer in sensory than in motor nerves, pulses longer than 0.15 ms are selected when attempting to locate a purely sensory nerve by inducing paraesthesia.

The accuracy with which peripheral nerve stimulators deliver the selected current tends to deteriorate as the current decreases. Manufacturers often quote accuracy of typically 1–5%, for a current setting of 1 mA. Amplitudes used for peripheral nerve localization, however, are less than 0.5 mA, where variability may be more than 80%.¹² A stimulator which delivers less than the set current may increase the risk of nerve damage during peripheral nerve block – with one that delivers more, a failed block may be more likely.

Nerve blocks carry a small risk of intraneural local anaesthetic injection with concomitant damage. Some nerve stimulators currently under evaluation calculate the electrical impedance (EI) of the circuit through the patient. This might be used to reduce the risk of intraneural injection, EI being higher when the stimulating needle is intraneural.¹³ The exact mechanism for this finding is yet to be elucidated and is controversial.¹⁴

Transient pacemaker inhibition by peripheral nerve stimulators has been reported,¹⁵ as has interference with the display of pacemaker impulses on the ECG monitor.¹⁶

Assessment of neural integrity

Peripheral/cranial nerves

Nerve stimulators may be used intraoperatively to help assess the integrity of peripheral nerves. In many instances the distal contractile response to a proximal stimulus, applied to an intact motor nerve, may be assessed by palpation. Stimulation of motor nerves may be useful to confirm neural function in surgery around the facial nerve, or in thyroidectomy to identify the recurrent laryngeal nerve. For the latter, electrode-embedded endotracheal tubes with attached electromyographic monitors have been employed with variable success.¹⁷

Spinal cord/nerve roots

Somatosensory evoked potentials

The incidence of neurological damage, secondary to traction and ischaemia of the spinal cord during scoliosis correction is approximately 0.5%. This may be reduced by monitoring somatosensory evoked potentials (SSEP).¹⁸ The monitors are complex and require an experienced neurophysiologist. Typically a pair of stimulating electrodes is placed over each posterior tibial nerve at the ankles for lumbar surgery, or over the median nerves for cervical surgery. Stimuli are administered at about 30 mA and 5 Hz. Recording electrodes may be applied at various points adjacent to the ascending tracts and proximal to the site of surgery. Usually two or more scalp electrodes are used (e.g. one frontal and one cervical), together with a reference and a ground electrode. The recorded responses are passed through a digital signal converter and a band-pass filter (20–1000 Hz). An average response is calculated from as many as 200 individual sweeps and the result displayed on the monitor screen. The period of interest is the first 100 ms after the stimulus, during which a characteristic W-shaped potential is seen in recordings from cortical electrodes (Fig. 17.4). Response latency also depends on the distance between the point of stimulation and the recording electrode.

Figure 17.4 Somatosensory evoked response. The stimulating electrode is over the median nerve. Recording electrodes are over A. the somatosensory cortex and B. the seventh vertebra. The sequential peaks and troughs in the evoked response are named according to their latency (ms) and direction. Particular periods of the response correspond to specific neuro-anatomical regions as shown.

(Thornton C, Sharpe RM (1998) Evoked responses in anaesthesia. British Journal of Anaesthesia 81: 771–781, © The Board of Management and Trustees of the British Journal of Anaesthesia. Reproduced by permission of Oxford University Press/British Journal of Anaesthesia.)

A baseline SSEP is recorded before induction of anaesthesia. If subsequent intraoperative spinal manipulation completely prevents conduction of the ascending impulse, the SSEP disappears altogether. When conduction is merely impaired, an increase in SSEP latency and a decrease in amplitude are observed. A 50% increase in latency compared to baseline would be regarded as a critical change and the operator would inform the surgeon accordingly. Surgical disruption apart, a number of factors affect SSEPs, including volatile anaesthetic agents, nitrous oxide, hypothermia, hypoxia and hypotension.

Motor-evoked potentials (MEPs)

MEP monitoring is now employed routinely in some spinal surgery centres, as SSEPs may have limited ability to detect ischaemia secondary to anterior spinal artery hypoperfusion. For spinal surgery, transcranial electrical stimulation is most commonly used. Though this carries a risk of injury secondary to induced contraction of the mandibular muscles; in practice serious complications are rarely seen. Alternatives include transcranial magnetic stimulation or direct stimulation of the rostral spinal cord.¹⁹ Multiple-pulse stimulations are performed and the results recorded at subcutaneous or intramuscular needle electrodes in arm and leg muscles. Unlike SSEPs, results can be available within 1 min.

As with SSEPs, the technician can inform the surgeon if MEPs disappear or if the stimulus required to elicit a response exceeds a pre-agreed threshold, indicating potential nerve damage. Spontaneous EMG activity consequent on stretch or compression of nerve roots, which can be detected by MEP recording electrodes, may provide useful information even in the absence of deliberate stimulation.²⁰

Monitoring ‘depth of anaesthesia’

For decades, anaesthetists have sought a monitor which might reflect the conscious level of patients undergoing general anaesthesia. A number of different variables, which may relate to consciousness, can be derived from the electrical activity of the cerebral cortex. Ideally, in variables of this sort, the range of values seen in the conscious state should not overlap with that seen in the unconscious state (i.e. a cut-off value would exist which is 100% sensitive and specific for consciousness). Furthermore, ideally any cut-off value should not be affected by patient physiology or the choice of anaesthetic agent.²¹

At present, the variables derived by available monitors do not meet these ideals. Current monitors generally work on one of two principles:

1. Spontaneous electroencephalography (EEG)

2. Provoked EEG in the form of auditory evoked potentials (AEP).

EEG

General principles, signal processing and artefact rejection

The EEG represents current flow in the cortical extracellular fluid, which is the result of post-synaptic potentials in cortical neurons. It is acquired through scalp electrodes. In anaesthetic monitors, between one and four electrodes are usually used, together with a reference electrode. Impedance should be kept to a minimum (i.e. less than 5 kΩ). EEG voltage is measured as the potential difference between two electrodes.

The complex waveform of the EEG comprises many individual sine waves, whose frequencies lie from zero to approximately 50 Hz (Fig. 17.5). Classically these are grouped into frequency bands (Table 17.1). For the purposes of analysis, the EEG is split into epochs of 1–4 s.

Figure 17.5 A. A 2-s epoch from an unprocessed EEG. Also pictured are some of its constituent sine waves, whose respective amplitudes, phase angles and frequencies are B. 30 µV, 60° and 20 Hz, C. 15 µV, 30° and 10 Hz, D. 4 µV, 15° and 4 Hz, E. 40 µV, 25 and 9 Hz.

Table 17.1 Electroencephalographic frequency bands

BAND NAME	APPROXIMATE FREQUENCY RANGE (Hz)
Alpha	8–13
Beta	13–30
Theta	4–8
Delta	<4

In order to allow processing, the EEG is converted from a smooth continuous analogue signal into a digital one. The fidelity of this conversion depends on the degree of resolution for both voltage and time. The greater the number of bits used, the smaller the change in voltage that can be translated from analogue to digital. EEG monitors usually use 12–16 bits of resolution. The frequency at which the analogue signal is sampled is also important. Too slow a sampling frequency fails to take account of the fastest sine waves and results in ‘aliasing’, where the digital signal incorrectly identifies a low-frequency waveform (Fig. 17.6). EEG monitors usually sample at frequencies above 250 Hz. Signal processing then enables rejection of artefact due to the electrocardiogram, ocular movement or mains interference.

Figure 17.6 Aliasing. A high frequency signal at 20 Hz is incorrectly identified as a lower frequency signal at 2 Hz because the signal is sampled at only 6 Hz.

Time and frequency domain analysis and data presentation

The EEG is then subjected to time-domain and frequency-domain analysis. The most widely used example of the former is the calculation of a burst suppression ratio (BSR). In deeply anaesthetized patients, the EEG may consist of isoelectric periods (suppression) interspersed with bursts of normal activity. The BSR is the fraction of time in an epoch in which the EEG is suppressed and is usually averaged over 60 s.

In frequency-domain analysis, each individual epoch is subjected to fast Fourier transformation. This mathematical process breaks down the EEG waveform into its constituent sine waves, from which a power (µV²) vs frequency (Hz) histogram can be derived. On many monitors, the frequency histograms from sequential epochs can be presented as a compressed spectral array (CSA) or a density spectral array (DSA) (Fig. 17.7).

Figure 17.7 Spectral array displays. These are created by fast Fourier transformation of sequential epochs of raw EEG signal. Data from each epoch are transformed to produce a power versus frequency histogram. In the compressed spectral array (CSA) above, histograms derived from sequential epochs are plotted in pseudo-three-dimensional fashion. The density spectral array shown is derived from the same EEG data as the CSA. Each histogram is represented in greyscale, with larger values depicted by darker shades. In each case, data from the most recent epoch are added to the bottom of the display.

(Rampil IJ (1998) A primer for EEG signal processing. Anesthesiology 89: 980–1002, with permission.)

For the purposes of objective comparison, the frequency spectrum may be represented by a summary variable. Two commonly used variables are the median frequency (MF), which divides the power in the spectrum into two equal halves, and the 95% spectral edge frequency (SEF), below which 95% of the spectral power lies (Fig. 17.8). As anaesthesia deepens, lower EEG frequencies generally predominate and there is a concomitant fall in MF and SEF. Despite this broad relationship, MF and SEF are unsatisfactory measures of consciousness during anaesthesia. They have not been related to clinical endpoints, nor are they independent of anaesthetic agents used.

Figure 17.8 The EEG frequency spectrum and derived variables.

Bispectral analysis and the EEG bispectral index

The bispectral index (BIS) is an EEG-derived variable which is calculated and displayed by several devices manufactured by Aspect Medical Systems: such as the BIS VISTA monitoring system (Fig. 17.9) and BIS modules in multimodal monitors produced by other manufacturers. Bispectral analysis of the EEG signal, a method which is integral to the calculation of BIS, is a method of addressing relationships among signal constituents in the EEG. The use of MF and SEF as indices of anaesthesia described previously relies on assumptions about the linearity of EEG data. In linear systems, there is a simple proportional relationship between cause and effect. However, this situation is demonstrably not the case for many natural phenomena, including the EEG where different frequencies within the signal may not be independent of each other. Bispectral analysis, a higher order analysis based on the phase relationships of individual frequency components to each other, goes some way to addressing this issue and has long been used to examine wave patterns in oceanography. In the EEG, it is particularly applicable to the relationships between the constituent sine waves (described above under frequency domain analysis). These relationships – ‘phase coupling’ – have some relevance to the EEG effects of anaesthesia, becoming progressively more pronounced with increasing anaesthetic dose.

Figure 17.9 Annotated diagram of the display panel: Aspect BIS VISTA monitor.

Image courtesy of Covidien, Ireland.

BIS is calculated from the EEG using a proprietary algorithm, which incorporates data both from bispectral analysis and more traditional methods. The constituent elements are:

1. two time-domain features:

a. BSR, see above

b. QUAZI suppression index reflecting the proportion of near (or quasi) isoelectric activity

2. two frequency-domain features:

a. ‘SynchFastSlow’ (from bispectral analysis – the log ratio of bispectral power in the 0.5–47 Hz and 40–47 Hz frequency ranges)

b. ‘BetaRatio’ (the log ratio of power in the 30–47 Hz and 11–20 Hz frequency ranges).²²

The BetaRatio constituent relates to a phenomenon called ‘beta activation’, a paradoxical increase in relative beta power occurring at low brain concentrations of some anaesthetics. The overall product of the BIS algorithm is a dimensionless scale from 1 to 100, which indicates the likelihood of consciousness. During general anaesthesia, a BIS of less than 60 is said to indicate a negligible chance of recall.

The development of the BIS algorithm, and of bispectral index monitoring in general, has been described in several reviews.^22,^23,²⁴ The first algorithm, BIS 1.0, was released in 1992. It was constructed using EEG data acquired from young, fit patients under anaesthesia. Originally, the EEG features that best predicted movement on incision were identified and combined in an index which was then tested prospectively. Research in the early 1990s, however, indicated that the neural mechanisms underlying the movement response to incision differed from those concerned with hypnosis. As a result, BIS 2.0 was formulated using the existing database, to predict hypnotic rather than motor endpoints.

By version 3.0 (1995), the database had increased to about 1500 anaesthetics. This version, and subsequent ones, has represented attempts to improve performance of BIS at very deep or light levels of anaesthesia and to enhance removal of electromyographic or electrocautery-related artefact. To this end, the latest VISTA generation of BIS monitors use the 4.0 or 4.1 versions of the algorithm. In these versions BIS is calculated using data from sensors containing four rather than three electrodes, such as the Quatro (Fig. 17.10), Extend, and semi-re-usable sensors (all Aspect Medical Systems), in which the additional electrode is placed above and lateral to the eyebrow. A bilateral sensor has also been released recently, though its advantages are unclear at present.

Figure 17.10 The BIS QUATRO sensor for use with the bispectral index monitor.

Image courtesy of Covidien, Ireland.

While BIS is now widely used, its reliability as an indicator of hypnosis during general anaesthesia depends on the agent. It is ineffective during ketamine anaesthesia and doubts remain over the precise relationship between BIS and clinical state during nitrous oxide sedation and high-dose opioid anaesthesia. In recent years, authors have suggested that the BIS algorithm is unnecessarily complex. They argue that a variable, analogous to SynchFastSlow, can be calculated from the EEG without bispectral analysis. This variable, PowerFastSlow, requires fewer data for its calculation. It too measures phase coupling, using a phenomenon known as bicoherence. Unlike the bispectrum, this is independent of signal amplitudes. In patients receiving a propofol, alfentanil and isoflurane anaesthetic, PowerFastSlow appears to predict the anaesthetized state as well as SynchFastSlow.²⁵

Other spontaneous EEG processing devices

Other monitors which record and process the EEG are available. The SEDLine (Hospira), formerly known as the Patient State Analyzer PSA-4000, was developed in a broadly similar fashion to the BIS monitor, using EEG databases and clinical correlates. Most of the EEG descriptors used in the algorithm for Patient State Index (PSI) calculation, relate to EEG power, though the suppression ratio is also taken into account. Six electrodes are placed on the forehead, to acquire data from four EEG channels. The resulting PSI ranges from 0 to 100 and PSI 25–50 is recommended for surgical anaesthesia.²⁶

In the Narcotrend monitor (MonitorTechnik), the EEG is acquired from one reference and two recording electrodes on the forehead. Artefact is rejected, the data are analyzed and an algorithm applied in order to assign a Narcotrend stage. There are six such stages – A (awake) to F (general anaesthesia with increasing burst suppression). A recent version also includes a numerical index. The algorithm is distinct from the BIS and PSI algorithms in that it was developed purely as a means for objective analysis of the EEG waveform, using time and frequency domain information. No clinical correlates were involved.²⁷

Effective intraoperative titration of general anaesthetic has been demonstrated using all three monitors described above, and the M-Entropy module (see below), with monitored patients recovering more quickly than controls.

Other commercial monitors based on the spontaneous EEG included the Cerebral State Monitor (Danmeter A/S). This is a small battery-operated handheld device attached by electrodes to the forehead. Both proprietary and routine ECG electrodes can be used. It acquires data from a single EEG channel. Fuzzy logic (if x and y, then z) is then applied to four calculated EEG parameters, namely alpha ratio, beta ratio, the difference between these ratios and burst suppression ratio, to determine the Cerebral State Index (CSI). The CSI is a dimensionless number scaled from 0 to 100.²⁸ Danmeter A/S ceased trading in 2008, although the products are still in circulation.

The SNAP II (Everest Biomedical Instruments) is based on a personal digital assistant, onto which can be ‘snapped’ a monitoring module. The single-channel EEG is acquired from a forehead electrode strip and the SNAP II index is calculated using low (0.1–18 Hz) and high (80–420 Hz) frequency components of the EEG. The index runs from zero to 100, the latter indicating the patient is fully awake. A SNAP index of 50–65 is recommended for general anaesthesia.²⁹

Entropy

In 2003, Datex-Ohmeda released a depth of anesthesia module for their S/5 anaesthesia monitor (Fig. 17.11) based on the entropy of the spontaneous EEG. If the awake EEG is characterized by a chaotic signal, then decreasing levels of consciousness are associated with a less disordered signal, as the number of signal generators diminish and slower wave activity becomes more dominant. By calculating the amount of disorder in the power spectrum of the EEG signal, it is suggested that anaesthetic depth may be objectively estimated. In contrast to the often opaque proprietary algorithms above, this system uses a relatively simple mathematical calculation of Shannon entropy with the resultant numerical scale deliberately manipulated to correlate with the now familiar BIS values. A three electrode sensor similar in appearance and application to that for BIS is used to acquire a single-channel EEG. From this, the module calculates two values: one between 0 and 91 for ‘state entropy’ (SE), which reflects cortical activity over the frequency range 0.8–32 Hz, and a second between 0 and 100 for ‘response entropy’ (RE) over the frequency range 0.8–47 Hz, to include components of the frontalis muscle electromyogram (EMG).

Figure 17.11 Datex–Ohmeda M Entropy module and electrode array.

The manufacturers recommend that both values should be between 40 and 60 during surgery under general anaesthesia. RE becomes equal to SE when the EMG power is zero, otherwise it is always higher than SE and rapid rises in RE are said to reflect relative analgesic inadequacy. Where SE is above 60, a higher anaesthetic dose is required. Where SE is satisfactory, but RE is more than 5–10 units above the SE value, more analgesic is needed. It is hypothesized that, in the case of deficient analgesia in a non-paralyzed patient, EMG facial activity increases before any change in the EEG activity, leading to an increase in RE before any change in SE.

Although global correlation with BIS appears good, agreement is poor. This may reflect differences in scale, despite numerical adjustment. In general, the M-Entropy module appears to perform no better than BIS in terms of correlation with effect site drug concentration and prediction of clinical endpoints. In sedated patients the frontalis EMG renders interpretation of SE and RE values difficult.³⁰

Auditory evoked potentials

Loss of consciousness under general anaesthesia is accompanied by changes in the electrical response of the cerebral cortex to an auditory stimulus. The period of particular interest is the early cortical response, illustrated in Fig. 17.12, which occurs approximately 10–100 ms after the stimulus (the mid-latency auditory evoked potential, or MLAEP). MLAEP waves are generated in the medial geniculate and primary auditory cortex. Anaesthesia increases the latency and decreases the amplitude of MLAEP waves. Threshold values for both Na and Pb latencies have been proposed as indicators of unconsciousness during anaesthesia. More complex MLAEP-derived indices, which better reflect its overall morphology, have potential clinical advantages and may feature in future monitors.³¹ Particular frequency components within the MLAEP may also have a role.³²

Figure 17.12 The auditory evoked response with its anatomical basis.

The A-Line AEP monitor/2 (Danmeter) is a hybrid spontaneous EEG/AEP monitor (Fig. 17.13). For the AEP component 65–70 dB click stimuli are administered at 9 Hz through headphones, the response being recorded with scalp electrodes. A ‘Click Detection’ function alerts the user if the auditory stimulus ceases for some reason. The response signal undergoes pre-processing, during which artefact is rejected and band-pass filtering applied. Subsequent extraction of the AEP from background cortical activity takes time and results in some latency. In current versions of the monitor, the replacement of ‘moving time averaging’ with an ‘autoregressive model’ (using a proprietary mathematical method called ARX modelling) has reduced the update delay time from 35 to 6 s. The latest A-line ARX Index (AAI) is the sum of absolute differences in the 20–80 ms window of the AEP and is a unitless index ranging from 0 to 99. This is preferentially derived from the AEP, but the EEG signal is used if AEP values are weak. This device is no longer in production.

Figure 17.13 The Danmeter A-Line AEP monitor: (1) recording scalp electrodes; (2) headphones.

aepEX monitor

Manufactured by Medical Device Management Ltd, this small handheld monitor, which is also available as a plug-in unit for modular monitors, generates a single number between 0 and 100 based on analysis of auditory evoked potentials (Fig. 17.14). The aepEX index, previously called the ‘AEP index’ and the ‘level of arousal score’, reflects the morphology of the MLAEP curves. It is calculated as the sum of the square root of the absolute difference between every two successive 0.56 ms segments of the AEP waveform.³³ The AEP curves recorded from scalp electrodes are produced by averaging 256 sweeps of 144 ms following the presentation of 7 Hz auditory clicks of 1 ms duration at 70 dB above normal hearing threshold. Stimuli are administered through small earphones with single-use gels, available in three sizes. The EEG signal is detected by three disposable sensors: two on the forehead and one on the mastoid process. By using a moving time averaging technique results are updated at 1 s intervals. Typical values in a conscious patient are 65–85 and, during surgery, 30–45.

Figure 17.14 The aepEX monitor together with sensors and earphones.

Photograph courtesy of Medical Device Management Ltd. UK.

The aepEX is said by the manufacturers to work equally well with all anaesthetic agents including ketamine and nitrous oxide. There is some evidence to support this particularly where ketamine is given in association with remifentanil.^34,³⁵

Clinical use of depth of anaesthesia monitoring

Of the above technologies, at present BIS has European and US market dominance and the largest volume of supporting evidence. AEP index and Entropy also appear to show efficacy in attempting to detect unconsciousness. In a direct experimental comparison during propofol sedation, an AEP-derived variable (the AEP index) was better than BIS at distinguishing the conscious from the unconscious state.³³ However, results from a large-scale trial indicate that BIS monitoring can reduce the incidence of intraoperative awareness in patients at risk of this complication,³⁶ although a trial comparing a BIS based protocol and one based on the measurement of end tidal anaesthetic agent showed no advantage.³⁷ At present, there is no corresponding evidence for the effectiveness of AAI-guided general anaesthesia and Danmeter, the manufacturer of both AEP/2 and the Cerebral State Monitor, has recently ceased trading.

New ways of using the information in the spontaneous EEG to reflect the hypnotic effect of anaesthetic agents appear frequently in the literature. In general, the drive is to develop a less complicated algorithm than BIS, while retaining its clinical utility. On most occasions, the reliability of the new technology is assessed in terms of its agreement with simultaneous BIS readings. In the more rigorous studies, the ability of new technology to predict clinical endpoints is compared with that of BIS.

Limitations

Non-anaesthetic factors affecting cerebral metabolism, e.g. hypotension, hypoxaemia, hypothermia and hypoglycaemia, can affect EEG-derived indices of ‘depth of anaesthesia’. The precise contribution of these factors to an anaesthetic effect, as compared to the output of the monitors, is not yet elucidated. Additionally, opiates, especially at high dose, have a particular effect on the EEG whilst being devoid of the ability to prevent recall. Other drugs: psychotherapeutic agents, ketamine and nitrous oxide may introduce further difficulties in producing inconsistent effects on the measured entropy of the EEG or producing an anaesthetic effect not legible to BIS.

It is important in using all monitors of depth of anaesthesia that we do not conflate the different questions that may be asked:

• Is the patient asleep or awake at present?

• How much effective anaesthetic agent is active in the patient?

• What will be the effect of surgical stimulation on arousal?

• Will there be movement in response to stimulation?

• Is there or will there be information processing or retention taking place?

The EEG analysis technique used by a depth of anaesthesia monitor, the patient physiological state and the particular anaesthetic agents in use will govern the veracity with which each of these questions may be answered.

It is conceivable that in due course we will decide that spontaneous EEG-based monitors in assessing current cortical activity will always tend to lack the predictive potential of monitors based on provoked EEG, which by their nature are assessing the ability of the brain to respond to a stimulus.

Pain measurement

In an attempt to facilitate detection of pain in patients unable to communicate reliably, a skin conductance algesimeter, the MEDSTORM AS 2005 monitor (Med-Storm Innovation), has recently been developed. Nociceptive stimuli increase sympathetic stimulation to skin sweat glands, thereby raising electrical conductance. The monitor acquires data from three palmar electrodes. The number of fluctuations per second in skin conductance (NFSC) appears to be a more useful variable than mean skin conductance, which is highly dependent on electrode position and type.

In one prospective study of the monitor in postoperative adult patients, NFSC correlated with pain score. Using a predetermined cut-off value for NFSC of 0.1, the monitor distinguished between those with no/mild pain and those with moderate/severe pain with a sensitivity and specificity of 88.5% and 67.7% respectively.³⁸ Though such a monitor might be helpful in some circumstances, the skin conductance method appears to be affected by anticholinergic agents.³⁹

Assessment of cerebral blood flow

During carotid endarterectomy (CEA), the adequacy of cerebral blood flow (CBF) may be assessed using a number of monitoring techniques. Their individual sensitivity and specificity in the detection of cerebral ischaemia, and their role in the prevention of adverse neurological sequelae, is disputed. Some of these techniques may have other surgical applications, for example, in cardiac or neurosurgery.

Methods include:

• Stump pressure measurement

• Transcranial Doppler ultrasonography (TCD (see below))

• EEG

• SSEPs (somatosensory evoked potentials)

• Intraoperative jugular bulb oximetry (neurosurgery only)

• Cerebral oximetry (predominantly cardiac surgery).

Stump pressure

Immediately after clamping the carotid artery the surgeon may insert, distal to the clamp, a needle attached to a standard pressure transducer. In principle, the adequacy of collateral blood supply from the circle of Willis may be reflected in the pressure at this point. Typically, a shunt might be inserted for the duration of clamping if the stump pressure is less than 50 mmHg. There are a number of problems with stump pressure. It is a one-off measurement, it does not consistently reflect flow, and the relationship between stump pressure and ischaemic EEG changes is weak. Nevertheless, in series where stump pressure is used to dictate shunt insertion, rates of shunt insertion are no higher than in series where more complex and expensive methods are used.⁴⁰

Transcranial Doppler technique (TCD)

TCD is probably the most widely used continuous technique and requires an experienced operator. The principles of Doppler ultrasonography are covered in Chapter 31. Traditional 5–10 MHz ultrasound frequencies do not penetrate the skull and TCD usually involves pulses at 1–2 MHz, with the probe placed over the thin bone of the temporal region (Fig. 17.15). At these relatively low frequencies, spatial resolution is poor and, therefore, the ultrasound is primarily used for measuring blood flow velocities. Even so, up to 30% of elderly patients cannot be insonated, even at the highest available energy output of the devices.

Figure 17.15 Transcranial Doppler ultrasound probe applicator for Sonara TCD system.

Image courtesy of CareFusion (UK).

During CEA, TCD is generally used to assess flow in the middle cerebral artery, which is found at a depth of about 50 mm from the temporal window. The probe emits ultrasound pulses and receives reflected frequencies; the frequency attenuation (Doppler shift) is used to produce a moving graph of flow velocity (cm s⁻¹) vs time. The waveform is characterized by its peak systolic velocity and its time averaged mean maximal velocity (Vmax). Vmax is typically 35–90 cm s⁻¹. Another useful variable is the pulsatility index, which reflects the resistance to flow in the examined artery. A TCD monitor is illustrated in Fig. 17.16.

Figure 17.16 Sonara Transcranial Doppler ultrasound monitor.

Image courtesy of CareFusion (UK).

Though changes in Vmax are widely regarded as being more significant than absolute values, it is unclear what constitutes a critical reduction. In one large study, ischaemia after clamping was considered severe if Vmax fell to 0–15% of baseline, mild if 16–40% and absent if greater than 40%.⁴¹ A significant decrease on clamping may influence the surgical decision to shunt. Subsequent Vmax changes may allow detection of shunt occlusion and intraoperative emboli, the latter producing a characteristic sound.

As well as decreased CBF during surgery, TCD allows detection of increased flow postoperatively in hyper-perfusion syndrome. This complication occurs in about 1% of patients undergoing CEA. Prompt recognition and administration of appropriate therapy may reduce the risk of cerebral haemorrhage.

Though TCD monitoring has been attempted in cerebral aneurysm surgery and resection of intracranial arteriovenous malformations, practical difficulties limit its usefulness.

SSEPs

Cortical SSEPs are affected by large reductions in CBF. Median nerves may be stimulated bilaterally and evoked potentials recorded at scalp electrodes as described above. A baseline response is recorded before clamping and again about a minute after. Subsequent recording at regular intervals is recommended as delayed changes are occasionally seen. SSEPs have been used successfully as a basis for selective shunting in CEA.⁴² On carotid clamping, a decrease in N20/P25 amplitude of greater than 50% from baseline has been proposed as an indicator of cortical dysfunction.⁴³

EEG

Cerebral ischaemia occurring on carotid clamping may induce changes in ipsilateral EEG frequencies and amplitudes. Raw EEG monitoring during carotid endarterectomy (CEA) has been advocated, although data interpretation requires considerable expertise. Processed EEG monitors may be more practical, but BIS appears unreliable for detection of cortical ischaemia ⁴⁴ and the predictive sensitivity of intraoperative EEG monitoring for immediate postoperative neurological deficit is poor.⁴⁵

Jugular bulb oximetry

More usually used in the intensive care unit, jugular bulb oximetry also allows intraoperative assessment of cerebral oxygen extraction during cerebral aneurysm, tumour and haematoma surgery.⁴⁶ An oximetric catheter, e.g. Opticath (Abbott), is introduced into the internal jugular vein cranially until the jugular venous bulb is encountered. Jugular venous oxygen saturation (SjvO₂) reflects the oxygen ‘supply:demand’ ratio and is usually 60–70%. Normal values give limited reassurance as they do not exclude focal cerebral ischaemia. An SjvO₂ of 90% or more is seen in hyperaemia – at less than 50%, SjvO₂ indicates increased oxygen extraction and impending ischaemia.