Almost all monitoring equipment is now controlled by microprocessors, making it possible to design equipment that needs minimal user calibration, offers self-diagnostics for fault conditions and also almost endless variability in user configuration of parameters.¹¹ Single modality monitoring is becoming increasingly rare as a result of the flexibility and miniaturization possible with microprocessor control. Even greater user variability is offered in integrated monitors (one display for several variables) by the use of a modular system, in which the module for each monitored modality may be replaced during use (hot swapped) in the event of a fault or to perform a different function.

The remainder of this chapter will cover non-invasive monitoring, whatever the physiological system being monitored or the source of the electrical signal being processed. Chapter 15 covers gas monitoring and Chapters 16 and 17 cover other aspects of cardiovascular and neurophysiological monitoring.

Monitoring biological electrical potentials

All living cells have a potential difference between the outside and the inside of the cell membrane, which generates small electric currents within and between cells. Such a potential difference, when associated with excitable cells acting in unison, such as muscle or nerve, generates a current sufficient to be detected by a measurement system. Depending on the origin of the biological potential, there is a wide range of amplitudes and frequencies to be measured, processed and displayed, as shown in Table 14.1.

Table 14.1 The range of amplitudes and frequencies generated by biological potentials from different sources

SOURCE OF BIOLOGICAL ELECTRICAL POTENTIAL	AMPLITUDES	BANDWIDTH (Hz)
ECG	0.5–4.0 mV	0.01–250
EEG	5–300 µV	DC-150
EMG	0.1–5.0 mV	DC-10000

The quality of the electronics within these monitors must, therefore, be of a high order to measure accurately the currents produced by such small electrical potentials within a wide range of frequencies. Amplifiers used to process biological electrical potentials should have the following properties:¹²

• The signal to noise ratio of the amplifiers should be high. This is the ability of an amplifier to ensure preferential amplification of the signal being measured in comparison to any electrical noise interfering with this process.

• The common mode rejection ratio should be high; this is achieved by having two input ports, one of which inverts the input signal, so that any random noise signals, such as inductively or capacitatively coupled signals, which are common to both ports, cancel each other out before entering the amplifier, while signal inputs, which are not common to both inputs to the same extent (such as the biological potential itself), are allowed to enter the amplifier. The difference between the two inputs is then subjected to electronic amplification. Fig. 14.4 shows single and double input amplifiers:

Where there are several amplifiers in a system, the input impedance of any single amplifier should be as high as possible so that the amplifier itself does not draw too much current from that being measured, thereby reducing its value.

The output impedance of an amplifier should be low so that the partially processed signal can be passed to the next stage with minimum attenuation.

In order to process preferentially that component of the signal within the appropriate range of frequencies without attenuation or distortion and to eliminate inappropriate frequencies, the amplifier should have appropriate filtering in its circuitry to give adequate bandwidth.

Figure 14.4 A. The output of a single input amplifier is adversely affected by stray signal coupling. B. The output of a differential input amplifier is relatively unaffected by stray signal coupling.

Even surface electrodes must be carefully designed in order to minimize degradation of the small surface electrical potentials. A further requirement of the monitors is high-quality electrical isolation between any parts touching the patient, such as the electrodes, and any other electrical components within the device, which might lead to an electrical hazard to the patient. This is particularly important, since it is the only situation where there is a deliberate electrical connection made between the patient and a device which might be connected to a high-voltage mains frequency source. If two pieces of electrical equipment are attached to a patient and one develops a fault allowing an unwanted earth pathway to occur, there is a risk of electrocution or diathermy burns. This is discussed in detail in Chapter 23.

The section below covers the electrocardiogram and monitoring of the neuromuscular junction, while Chapter 17 covers the electroencephalogram and its derivatives.

The electrocardiograph (ECG)

Despite the increasing intraoperative use of other monitors of cardiac function, such as ultrasound (see Chapter 16), the electrocardiogram remains extremely useful at detecting ischaemic events, providing close attention is paid to lead selection, amplification and filtering.¹³ The electrical depolarization and repolarization of the myocardium is detectable on the surface of the skin, by the electrocardiogram in the form of the familiar PQRST complex. This can be recorded by the use of electrodes connected to limbs and chest, which look at the electrical vector of the ECG from slightly different points of view. The relationship between the electrical axis of the heart and its detection by different limb leads was described by Einthoven in 1901. The signal from the electrodes is then fed into an amplifier, which should meet the requirements discussed above.

As indicated in Table 14.1, the electrical potential from a surface ECG is in the range 0.5–4 mV, lying in the frequency range (bandwidth) of 0.01–250 Hz. The ECG is a complex waveform, which consists of a series of sinusoidal waves with different amplitudes, frequencies and phase relationships to each other. The ability of the monitor to process the ECG waveform depends on its ability to respond to the range of different frequencies of these sinusoidal components, with faithful, unattenuated and undistorted reproduction of the signal, and with the desired amplification or gain. This is a measure of the bandwidth of the monitor and the ECG monitor should ideally have the maximum required bandwidth indicated above. This is not practicable without also allowing amplification of noise within that bandwidth, thereby interfering with the ECG signal. Noise can originate from the amplifier circuit itself, from chest muscle electrical activity, from electromagnetic interference (inductive coupling), or from capacitative coupling to neighbouring electrical equipment such as diathermy apparatus (see Chapter 24). It is, therefore, more common to reduce the bandwidth to 0.05–100 Hz in a monitor with which the clinician wishes to make a range of diagnoses and to an even narrower bandwidth of 0.5–40 Hz in a monitor used in the operating room. The effect of this narrower bandwidth is to exclude high- and low-frequency components of electrical signal, which interfere with the ECG without adversely affecting its quality or the clinician’s ability to diagnose ischaemia and arrhythmias.

Fig. 14.5 shows graphically the effect of electronic filtering, which is achieved by the addition of appropriate electronic components to an amplifier circuit. A low pass filter allows low-frequency components to pass through the amplifier, blocking high-frequency components. A high pass filter does the converse, allowing through high-frequency components. A band pass filter allows through a range of frequencies, blocking signals of frequencies above and below this bandwidth. There is a significant DC voltage of up to 25 mV at the skin–electrode interface, partly due to the resistance in the layers of the skin and partly due to the electrolytic reaction between the Ag/AgCl gel and metal component of the electrode assembly. An ECG amplifier has to eliminate this DC voltage as well as the high-frequency noise and, therefore, a band pass filter is appropriate.

Figure 14.5 The effect of different electronic filters in allowing signals of different frequencies to pass through the amplifier.

Fig. 14.6 shows a block diagram of the components of a modern ECG monitor. As discussed above, the characteristic features include a differential input to ensure common mode rejection. Electrical isolation of the amplifier is also important for reasons discussed earlier, using either an isolation transformer with good insulation between primary and secondary windings, or an optical isolator, in which the output signal from the amplifier is converted by a light-emitting diode into a light signal and converted back into an electrical signal by a photo-detector, with good insulation between the two.

Figure 14.6 Components of a modern ECG monitor.

Even these design features may allow the ECG signal to be overwhelmed by high-power radio frequency diathermy signals. To deal with this, an additional technique is adaptive noise filtering. This is a digital electronic technique in which the noise signal is separately detected and digitally subtracted from the output signal, theoretically leaving a clean ECG signal. As Fig. 14.6 shows, analogue to digital conversion of the signal occurs to allow microprocessing of the signal. One of the many things a microprocessor provides is the storage, in digital form in the random access memory (RAM), of enough signal information to reproduce a screen width’s worth of signal. This produces a more persistent and more easily readable screen than the previous phosphor dot screen. The RAM allows continual updating of the stored information so that the most recent 5 or 10 s of data are available. At the same time, the complete RAM storage is rapidly read through the microprocessor and a digital to analogue converter to a display. Thus, a rolling display of the most recent information is available and can be frozen on the screen for closer examination. Modern ECGs have adjustable rate alarms.

The causes of erroneous traces on an ECG include inappropriately positioned electrodes, left to right misconnections, poor electrode contact, capacitative and inductive interference, and voluntary and involuntary movement, such as shivering and excessive muscle tone.

Neuromuscular monitoring

The neuromuscular junction is monitored in order to determine onset and recovery from pharmacological neuromuscular blockade.¹⁴ A supramaximal electrical stimulus of between 10 and 50 mA is applied near an accessible peripheral motor nerve (for example, the ulnar or facial nerve) and the response of the appropriate muscle group is assessed, either clinically, mechanically or electromyographically (see Chapter 17). The stimulus has a square waveform of width 0.2 ms, to avoid the nerve firing repeatedly. The electrical stimulus applied in a clinical situation is either a single twitch or, more commonly, a train of four twitches (TOF), delivered at 2 Hz,; or as a tetanic stimulus of duration not more than 5 s, delivered at 50 Hz. Double burst stimulation, which is two 40 ms tetanic bursts 750 ms apart, should also be available.

In non-depolarizing neuromuscular blockade, there is a characteristic fade on the muscle response to a TOF, and a characteristic temporary augmentation of the muscle response to further twitches delivered 3 s after a tetanic stimulus. The responses are distinguishable from either no neuromuscular blockade or a depolarizing block. Application of the stimulus is painful and should not be carried out on an awake patient.

The nerve stimulator should have a constant current output.¹⁵ This is desirable to ensure a constant current between the electrodes of the stimulator, whatever the changes in electrical impedance of the tissues to which they are applied. Many devices deliver constant current up to a tissue impedance of 2.5 kΩ. However, in some patients, the tissue resistance may be higher, causing a drop in current output, delivering a stimulus which is no longer supramaximal.

The electrodes may be of the type incorporating Ag/AgCl gel, used for ECG recording. Since a supramaximal response is required, an adequate current density is necessary at the electrodes, the surface area of which should not be too large.

The most common way of assessing the muscular response to a supramaximal nerve stimulus is to make a visual clinical assessment of the responding muscle (e.g. levator palpebrae superioris in the case of the facial nerve or adductor pollicis longus in the case of the ulnar nerve), which is usually adequate. If a mechanical assessment is made (mechanomyography), it is usually done on the adductor pollicis longus muscle in response to ulnar nerve stimulation. The arm is rigidly clamped and a force transducer is placed in contact with the thumb, correctly orientated, with a preload of 100–300 g applied to ensure isometric thumb contraction. The transducer converts the force of contraction of the thumb into an electrical signal, which can be processed and displayed.

Electromyographic (EMG) assessment in response to peripheral nerve stimulation is no longer used to assess pharmacologically induced block. It is, however, more common elsewhere than it was previously because of advanced signal processing and computer data handling techniques.¹⁶ In anaesthetic practice it is more likely now to be met as the technique used to monitor spinal cord integrity during surgery (see Motor-evoked potential monitoring, Chapter 17).

The EMG represents a collection of muscle action potentials which can be recorded through skin electrodes or, ideally, intramuscular needle electrodes. The evoked EMG signal, in response to nerve stimulation, is a high-speed event, with very high-frequency components, as Table 14.1 indicates, and requires considerable processing before any meaningful display.

Blood pressure monitoring

Monitoring blood pressure (BP) in the perioperative period has been routine since 1903 and is part of contemporary minimum monitoring standards. It is usually done non-invasively, except where there is a clinical need dictating ‘beat to beat’ invasive monitoring. Arguably, cardiac output would be a more useful measure of cardiovascular wellbeing, but has always been more complex to carry out than blood pressure measurement. Non-invasive blood pressure measurement became relatively easy to do early on in the history of clinical monitoring, but the earliest blood pressure measurements were invasive, using a manometer (see below). The manometer is still used in many settings, as a mercury manometer attached to a non-invasive sphygmomanometer cuff (see below) or as a water manometer attached to a central venous catheter to measure pressure directly. The manometer is a column of fluid, at the bottom of which the pressure is equal to the weight of fluid above it, divided by the cross-sectional area of the tube. Hence the pressure P is given by:

where ρ is fluid density, g is acceleration due to gravity, h is the height of fluid in the manometer column and A is the cross-sectional area of the tube. Hence the colloquial use of a column height (e.g. mmHg, cm H₂O) to describe pressure has evolved. The hydraulic pressure being measured is applied to the bottom of the manometer column or to one side of the ‘U-tube’ form of the manometer, shown in Fig. 14.7.

Figure 14.7 Two sorts of manometers to measure pressure: A. a U-tube differential manometer; B. a single tube manometer, of the type used in a sphygmomanometer.

The venous circulation is a low-pressure system, and in the upright or partially upright patient, the venous pressure at the head will be significantly less than at the feet; this should be remembered in deciding where to site the pressure transducer, when measuring central venous pressure. This difference in arterial pressure from top to bottom may not be so apparent, because the arterial circulation is a high-pressure system with regionally adjustable resistance to flow.

Non-invasive arterial blood pressure measurement techniques include auscultation using a sphygmomanometer, the oscillotonometer, the oscillometer and the volume clamp method (Peñaz technique, see below).

Non-invasive arterial blood pressure (NIBP) measurement

NIBP methods all function by inflating an occlusive cuff around a limb to a pressure above the expected arterial pressure and, on subsequent gradual, stepwise or continuous deflation,¹⁷ detecting the return of the pulse downstream of the cuff. Detection methods are described below in more detail. Most non-invasive methods are necessarily intermittent, although the volume clamp method of Peñaz (Finapres; see later) is continuous. In all non-invasive methods that use a cuff, the cuff size is important for accuracy; the width of the inflatable bladder of the cuff should be 40% of the mid-circumference of the limb concerned and its length should be twice this width. A cuff which is too narrow overestimates the blood pressure, while a cuff which is too wide or too loosely wrapped around the limb, underestimates it. Normally the cuff is wrapped round the upper arm to target the brachial artery, but if this is unsuitable (e.g. surgery on the arm or lymphoedema due to breast surgery), then the popliteal artery is targeted by wrapping the cuff around the calf, or, indeed, the posterior tibial artery if the cuff is wrapped around the ankle.¹⁸

The sphygmomanometer

The sphygmomanometer is a device that uses a combination of a pneumatic cuff (to wrap around the limb in which the arterial pressure is to be measured), an inflating bulb, a release valve, and a mercury manometer of the type shown in Fig. 14.7B. The classical method of pulse detection is auscultation. If the cuff is used on the upper arm, the sounds of the pulse returning are heard over the brachial artery. These sounds are known as Korotkoff sounds and their onset corresponds to the occurrence of turbulent flow in the artery as the cuff pressure falls below systolic blood pressure; the muffling or disappearance of these sounds corresponds to diastolic pressure.

The reason that Korotkoff sounds correspond to these pressures is unclear, but it is, therefore, not altogether surprising that it does not correlate well with invasive BP measurement under all circumstances. It has been shown that this method overestimates at low pressure and underestimates at high pressure, and has large inter-observer variation.¹⁹ On the other hand, the method is simple, uses low-level technology, and has a long history of use. Other methods of pulse detection with the sphygmomanometer include palpation and ultrasound. Palpation is more prone to error in the presence of bradycardia or too rapid cuff deflation and it has been shown also that this method underestimates systolic pressure by 25%.²⁰

The oscillotonometer

Fig. 14.8 shows the double cuff system used in oscillotonometry. The upper cuff performs the same function as that in the sphygmomanometer, namely limb occlusion. The lower cuff is wider and acts as the pulse detection system. Both cuffs are connected via an inflating bulb and air release valve to an airtight box containing two aneroid barometers. One aneroid (B₁) is relatively rigid and its inside is connected to atmosphere to measure the absolute pressure. The other aneroid (B₂) is relatively sensitive and its inside is connected to the distal cuff. Both aneroids are connected together by a lever, which is also toggled to the dial pointer. When the system is pressurized by manual inflation (using the bulb), both cuffs, the airtight box and the aneroid B₂ are filled to the same pressure. This causes the aneroid B₁ to be compressed and via its mechanical linkage moves a pointer to display the pressure in the system.

Figure 14.8 The working principle of an oscillotonometer.

Reproduced from Magee P, Tooley M (2005) The physics, clinical measurement and equipment of anaesthetic practice. By permission of Oxford University Press.

By switching on the release valve, a narrow diameter tube connecting the inside of aneroid B₂ to the atmosphere is opened and air is allowed to leak gradually out of the system. This allows deflation of the lower cuff to be marginally delayed behind the upper cuff. As the upper cuff pressure falls below systolic pressure, arterial pulsations impinge on the lower cuff and these are transmitted to aneroid B₂; with the release valve switch in this position, aneroid B₂ is also connected to the pointer and the pulsations, which are shown on the dial, increase significantly as systolic BP is reached. On releasing the valve switch, the pointer is reconnected to aneroid B₁ and the actual BP is indicated at this point. With the valve reswitched to air leak mode, the pulsation fluctuations gradually disappear as diastolic pressure is approached, and the actual pressure at which this occurs is once again read by reconnecting the pointer to aneroid B₁. Its accuracy is better at systolic pressure than diastolic.

Oscillometry

This method uses a single cuff that both compresses the limb and detects pulsations and uses hydraulics linked to modern electronics to obtain BP measurements. Some systems use a single tube for both inflation and detection, and others use two. Fig. 14.9 shows the components of the system. The hydraulic system consists of a pump to pressurize the system and a solenoid valve through which to depressurize the system, either in discrete steps or continuously, connected to a microprocessor. The detection system consists of a pressure transducer, whose output signal is fed through filters to amplifiers. On deflation, one part of the transducer signal goes through a high pass filter to pass only the high-frequency pulsation components to a high-gain amplifier that amplifies the pulsations due to the oscillations of the arterial wall.

Figure 14.9 The components and working principles of an oscillometer.

The onset of the rapid increase of pulsations corresponds to systolic pressure, maximum pulsations to mean pressure, and the rapid offset of pulsations to diastolic pressure. The other part of the transducer signal goes through a low-pass filter and a low-gain amplifier to produce a signal proportional to the cuff pressure. The outputs from both amplifiers are passed through an analogue to digital converter to the microprocessor, which calculates systolic, mean and diastolic pressures. Depending on the exact algorithm being used, the microprocessor may also compare the measured systolic and diastolic values to the values calculated from the mean value, at which the device has the greatest accuracy, as the oscillations are maximal at mean pressure. The microprocessor also controls the pump and solenoid valve and connects to a display.

Peñaz volume clamp technique

This method depends on the hypothesis that if the transmural pressure of an artery (the difference between the externally and internally applied pressure across the arterial wall) is kept constant, then the diameter of the artery also remains constant, as will the volume of blood within it and the absorption of infrared light across its lumen. The device consists of a low compliance finger cuff and tubing connected to a rapidly responding solenoid valve and air pump. It also has a light emitting diode as an infrared light source, which transmits light through the finger; the transmitted light is detected on the opposite side by a photodetector. The pump, the solenoid valve and the microprocessor function together to keep the light transmission through the finger constant by maintaining the transmural pressure. The device does this by altering the compression in the finger cuff throughout the cardiac cycle. At any instant the cuff pressure is the same as arterial pressure, and the low compliance of the system ensures rapid response. The output on the display therefore resembles an arterial trace. The Finapres, which is the manufacturer’s name given to the device, is shown schematically in Fig. 14.10. There have been reports of varying accuracy^21,²² and it is no longer in extensive clinical use.

Figure 14.10 The components of the ‘Finapres’ volumetric clamp method for measuring blood pressure.

Pulse oximetry

Principles

Table 14.2 shows that it is surprisingly difficult even for trained anaesthetists to ascertain accurately the patient’s state of oxygenation, particularly where the oxygen saturations have fallen below optimal levels. Arguably pulse oximetry, first introduced in the early 1980s, has revolutionized clinical monitoring in this respect. It should not be thought of as a replacement for other oxygen monitors, such as those on anaesthetic workstations, but it does provide the best non-invasive monitor of patient oxygenation. The pulse oximeter uses two technologies: one is pulse plethysmography to detect a pulse waveform; the other is infrared spectroscopy to detect the absorption, by the tissue under the probe, of light at two wavelengths in the red and infrared wavebands.

Table 14.2 The ability of experienced anaesthetists to detect hypoxaemia

Reproduced from Magee P. Tooley M (2005) The physics, clinical measurement and equipment of anaesthetic practice. By permission of Oxford University Press.

Pulse plethysmography detects the cyclical change in volume of the artery as change in light absorption across it, an increase in one being an increase in the other. Red or infrared light is used for the same reasons as in clinical spectroscopy, namely that there are absorption spectra in this waveband across an artery. The absorption signal is then electronically processed, including amplification and display. In a modern pulse oximeter, this signal processing and display scaling occurs automatically, so the size of the plethysmographic signal cannot be taken as a quantitative indicator of pulse volume; a pulse volume below a certain threshold may not be detected and amplified at all.

Apart from the objective indication of SpO₂, pulse oximetry may be described as tending to ‘fail-safe’, in that the technique fails and therefore alarms if the pulsatility of the pulse waveform decreases to below a critical level. Further information may be gained by the display of a plethysmographic trace, which enables the differentiation between the pulse waveform and artefacts. No pulse oximeter reading should be accepted unless the plethysmographic trace can confirm lack of artefact (see below, under Limitations).

The spectroscopic aspect of the technology is based on the absorption of light by a ‘dye’ in tissues, in which absorption is proportional to concentration of the dye and the path length through the tissue, and which also depends on the wavelength of light used; these factors are embedded in the Beer-Lambert laws (see Chapter 15). Fig. 14.11 shows the absorption spectra in the red and infrared wavebands of a number of haemoglobin species, including oxygenated and reduced haemoglobin, as well as the less common species, which can nevertheless unhelpfully contribute to a pulse oximetric signal.

Figure 14.11 The absorption spectra in the red and infrared region for different species of haemoglobin.

The pulse oximeter has light-emitting diodes (LEDs), which transmit light alternately at two wavelengths: 660 nm (red) and 940 nm (infrared). It can be seen that at 660 nm, the absorption of reduced haemoglobin exceeds that of oxygenated haemoglobin and at 940 nm, the converse is true. The absorption scale in Fig. 14.11 is logarithmic, so the differences in absorption are large. The device uses two wavelengths, because it needs to distinguish between absorption of light in pulsatile and non-pulsatile tissues under the probe. Ideally the two wavelengths used should correspond to those at which absorptions of both haemoglobin species (reduced and oxygenated) are equal (the isobestic point) and at which they are furthest apart. LEDs of 660 and 940 nm are used because of their ease of production, constant outputs and narrow bandwidths.

The probe is put on a digit, an earlobe,²³ the bridge of the nose or the forehead²⁴ and the more peripherally placed it is, the longer the delay in the device detecting an event causing oxygenation change.²⁵ Probes should be appropriately designed for paediatric use²⁶ because of the penumbra effect where, because of the small size of the digit to which the probe is attached, the path length of the light at the two wavelengths differs significantly. The LED’s rate of alternating on and off is 400 Hz, with a measurable ‘off’ phase, so the photodetector allows for ambient lighting. The photodetector measures the light transmission (transmission fraction = 1 – absorption fraction) of both pulsatile (AC) and non-pulsatile (DC) components at both wavelengths. The pulsatile components are assumed to come mainly from the arterial blood and non-pulsatile components from all other tissues including venous blood. The AC component of absorption represents 1–2% of the total, which puts significant demand on the accuracy of the device. Fig. 14.12A shows the raw signal, with different transmissions of each component at each wavelength. The first stage of signal processing is to change the amplitude of the signal at one of the wavelengths in order to equalize the DC components. The AC components, which are the signals of interest in this context, are now measureable against comparable denominators. The microprocessor in the device calculates the ratio of absorption of the AC component at 660 nm to the absorption of the AC component at 940 nm.

Figure 14.12 A. The crude AC and DC signals from a pulse oximeter probe are ‘normalized’ to make the DC components equal. B. The curve of absorption ratio to SpO₂.

Reproduced from Magee P, Tooley M (2005) The physics, clinical measurement and equipment of anaesthetic practice. By permission of Oxford University Press.

The microprocessor has stored in its memory a number of discrete values of oxygen saturation of arterial blood samples from healthy subjects, measured by a multi-wavelength co-oximeter. Fig. 14.12B shows this ratio plotted against the oxygen saturation and represents the calibration curve for the pulse oximeter. The term SpO₂ is used to describe the oxygen saturation derived from this curve. Note that an absorption ratio of 1.0 corresponds to a SpO₂ of 85%. The SpO₂ displayed is a value averaged over a number of beats, so a change in saturation may not immediately be displayed.

Limitations

Clearly, most values on the calibration curve shown in Fig. 14.12B are either interpolated or extrapolated. Since it would be unethical to collect blood from subjects who had been exposed to life-threatening hypoxia, it is to be expected that SpO₂ values below about 80% will be less accurate. In normal clinical circumstances, this is tolerable because values below 90% would be acted upon.

Pulse oximeters are designed to take account of two haemoglobin species, oxygenated and deoxygenated haemoglobin, to calculate functional saturation. As Fig. 14.11 shows, the presence of other haemoglobin species, such as methaemoglobin or carboxyhaemoglobin will affect the measured absorbances at the two wavelengths used. In particular, it can be seen that carboxyhaemoglobin resembles oxyhaemoglobin at 660 nm. This means that the pulse oximeter may give a falsely high SpO₂ in smokers, in whom COHb levels can reach 20%. Clearly a pulse oximeter should not be used to assess the oxygenation of a patient who has suffered from carbon monoxide poisoning. Similarly, methaemoglobin, caused by a number of drugs including local anaesthetics and nitrates, resembles deoxygenated haemoglobin at 660 nm. Only a co-oximeter with a minimum of four wavelengths can distinguish these four species, to calculate fractional saturation.

If vascular tone is markedly altered, then there is some limitation to the accuracy of pulse oximetery. This applies to hypertension or vasoconstriction induced by cold ²⁷ or otherwise;²⁸ however, the plethysmographic pulse width has been shown to reflect changes in vascular resistance,²⁹ and ear pulse oximetry may be more useful in peripheral vasoconstriction.²³ Where vascular tone is reduced, such as in hypovolaemia or sepsis, a peripheral plethysmographic signal may not be detectable.³⁰ However, the quality of the waveform³¹ may be useful in determining the clinical response to fluid resuscitation. If the venous circulation is hyperdynamic, as in tricuspid regurgitation,³² or if the perivascular tissue is pulsatile,³³ then the oximeter will tend to underestimate arterial oxygenation values.

Foetal haemoglobin has the same properties of light absorption as adult haemoglobin within the wavebands being discussed, so the pulse oximeter should be as accurate in neonates as adults. Bilirubin does not absorb light significantly in the waveband of interest; therefore, jaundice does not affect the accuracy of the pulse oximeter. Both foetal haemoglobin and bilirubin, however, affect the accuracy of a multi-wavelength co-oximeter. Skin pigmentation does not usually affect accuracy, but some dark nail polish does. Intravenous dyes, such as methylene blue and indocyanine green, alter the absorption spectrum of haemoglobin in the range of the pulse oximeter, thus attenuating its accuracy.

Pulse oximeters are also prone to error in the presence of movement and vibration,³⁴ or electromagnetic interference from ambient light, diathermy or mobile telephones.³

It is inappropriate to compare SpO₂ measured by a pulse oximeter with the value of oxygen saturation derived from the oxygen dissociation curve following measurement of pO₂, pCO₂ and pH.

Body temperature monitoring

At both physiological and biochemical levels, the human body functions optimally at 37°C. Anaesthesia and surgery both militate against this by tending to allow body temperature to fall and recovery to be delayed after prolonged surgery (see also Chapter 30). On the other hand, malignant hyperthermia is a potentially fatal condition caused by some anaesthetic drugs in patients pharmacogenetically predisposed to it. It is, therefore, essential for the anaesthetist to monitor body temperature.

A traditional way of measuring the temperature of a patient is to use a glass thermometer. This consists of a glass fluid-filled bulb attached to a calibrated glass tube. The glass bulb is placed against the tissue where temperature needs be measured, causing the fluid contained therein to heat up to the same temperature as the tissue. The resultant expansion of the fluid causes it to move into the calibrated glass tube as a column. The temperature can be read off the tube at the point where the head of the fluid column stops. A constriction is placed at the base of the tube so that when the bulb temperature drops and the bulb fluid contracts, the fluid column breaks allowing the final reading of the thermometer to be maintained.

Mercury is frequently the fluid used as its expansion characteristics allow it to cover a wide range of temperatures. Alcohol is also used, covering a narrower measurement range.

A thermistor is a semiconductor device whose electrical resistance changes with temperature in an almost linear fashion. It is the basis of both the nasopharyngeal temperature probe ³⁵ and some tympanic membrane thermometers.

A thermocouple consists of two dissimilar metals, connected together at the junction or measuring end and completing an electrical circuit (the opposite ends form the tail or reference end). A potential difference is generated between the two ends proportional to the difference in temperature between them, although the relationship is a non-linear one. This is the Seebeck effect, and is the basis of thermocouple function.

In the infrared tympanic thermometer (Fig. 14.13) a series of thermocouples (thermopile), detect the infrared radiation from the tympanic membrane.³⁶ The thermopile generates a potential difference proportional to tympanic membrane temperature.