Summary

The oxygen stores of the body are small, so life-threatening hypoxaemia can develop very rapidly with few clinical signs. The availability of robust and reliable pulse oximeters has revolutionised the safe monitoring of patients with unstable cardiorespiratory conditions, and those having medical and surgical procedures. While oximetry is now best practice in these circumstances, care must be taken in interpreting the results. There are confounding factors that may produce an erroneous signal and physiological factors that will affect the interpretation of the result. In the absence of these factors, the instruments are accurate detectors of arterial oxygen saturation, in the range between 100% and 70% with varying but reasonable performance down to 55%. The basic principles of operation are important to understand so that physiological interpretation is adequate and erroneous results can be identified.

Introduction

Patients at risk of hypoxaemia may need continuous monitoring of their oxygenation. Blood gas analysis requires arterial puncture and only measures the oxygenation at the time of the sample. By measuring oxygen saturation (instead of partial pressure) pulse oximetry enables non-invasive monitoring. The continuous measurement of the pulse rate is a bonus.

The technology supporting clinical oximetry has been available for more than 80 years, but pulse oximeters have only been commercially available for about 20 years. Early oximeters required a cumbersome heated probe to 'arterialise' blood in the ear lobe. They were also difficult to calibrate and notoriously unstable. Nowadays relatively cheap and reliable oximeters have revolutionised the in vivo monitoring of patients' oxygenation during a wide range of critical clinical situations.1

Physiological principles

While oximeters may be used to assess the efficiency of pulmonary gas exchange, at least in relation to oxygen uptake, they are more suited to assessing the adequacy of tissue oxygen delivery. Measuring oxygen in the arterial blood is important because serious acute hypoxaemia is notoriously difficult to detect clinically and by the time clinical cyanosis develops, the patient is usually in a parlous state. The oxygen stores of the body are small so the viability of many tissues is critically dependent on continuous delivery of an adequate oxygen supply. Oxygen delivery is proportional to the blood flow and arterial oxygen content (CaO2 [mL O2 per 100 mL blood]). For the whole body:

oxygen delivery = cardiac output (Q) x CaO2 x 10

These variables are difficult to measure directly and rapidly, however CaO2 is linearly related, at least over relatively short periods of time, to the saturation of haemoglobin in arterial blood (SaO2). As oximeters provide a rapid and reliable in vivo measure of SaO2 this variable can be substituted for CaO2. This has been a very valuable advance, as long as the principles and sources of error are understood.

Technical aspects

A pulse oximeter detects the change in transmission of two wavelengths of light across a capillary bed, usually in the finger. The sensor is placed on the nail with the light source against the finger pulp. The detectors can be small because they are only receiving two wavelengths, one to detect oxygenated haemoglobin (O2Hb) and one to detect reduced haemoglobin (HHb). The absorption of light is related to the expansion of the capillary bed with the pulse (Fig.1). By comparing the light transmission through the pulsatile 'arterialised' capillary blood with the non-pulsatile venous blood the oximeter can calculate the haemoglobin saturation. Saturation is calculated as:

O2Hb / [O2Hb + HHb]

This is the so-called 'functional saturation', and is expressed as a percentage.2

It has been suggested that pulse oximeters should be called 'pulse spectrophotometers'. This would emphasise that they are inferring oxygen saturation from the well-known colour change between oxygenated and reduced haemoglobin and that, despite the elegant use of the pulse form to separate arterial blood from the other light absorbing structures in the finger, there are sources of error inherent in this technique which need to be appreciated.

While the vast majority of devices in clinical use measure transmitted light, newer devices are being designed to measure the light reflected off pulsating tissue surfaces. These devices are being used in perinatal monitoring and in patients whose peripheral perfusion may be compromised, as in open heart surgery. Reflectance devices are currently hampered by poor signal-to-noise ratio and the need to detect very small pulsatile signals, but advances in technology are likely to overcome these difficulties.

Sources of error

The results of pulse oximetry can be affected by technical problems and physiological factors.

Calibration problems

The machines do not need regular calibration by the operator and the probes and electronics are extraordinarily robust. The machine will not display a result and will warn if it cannot detect an adequate pulse signal. It is fitted with an alarm, which can be set at low (or high) saturation levels as desired. Original calibration by the manufacturer is based on the empirical relation between in vivo pulse oximetry (SpO2) and the SaO2 measured on simultaneously sampled arterial blood in a CO-oximeter.

CO-oximeters, so called because they measure carboxy- or CO haemoglobin, are now fitted to all modern blood gas analysers. They use multiple wavelengths of light to detect the four different forms of haemoglobin. The calibration process generally relies on data generated from healthy volunteers made hypoxaemic to generate SaO2 values between 70% and 100%. When the SaO2 is reduced to between 70% and 40% the pulse oximeters become significantly inaccurate, particularly below 55%, and fail to track rapidly developing profound hypoxaemia.3However, it can be argued that the accurate detection of falls between 85% and 70% is of most use in clinical monitoring.

Physiological factors (Table 1)

The presence of abnormal haemoglobins disturbs the relation between SaO2 and CaO2. 'Functional saturation' ignores the possible presence of methaemoglobin, carboxyhaemoglobin and other abnormalities of haemoglobin. These abnormal forms will not carry oxygen normally and will add to the denominator of the O2Hb / [O2Hb + HHb] ratio. Usually abnormal haemoglobins only comprise a few percent of the total, even in heavy smokers who have increased concentrations of carboxyhaemoglobin. However, common drugs such as paracetamol and sulfa drugs can induce the formation of methaemoglobin. Anaemia will also reduce the oxygen content without changing the calculated functional saturation.

There will be difficulties relating the SpO2 to PaO2 if the position of the oxyhaemoglobin dissociation curve has been shifted by influences such as acid-base balance and carbon dioxide tension. If the SpO2 is above 92% the partial pressure of oxygen (PaO2) can change rapidly with very little change in saturation (Fig. 2). This latter physiological feature limits the usefulness of pulse oximetry in, for example, the assessment of pulmonary gas exchange efficiency and the detection of hyperoxia in preterm babies. Measurement of the partial pressure of arterial oxygen (PaO2) from in vitro samples or transcutaneous electrodes may be preferable for monitoring hyperoxia in the 'flat' part of the dissociation curve.

Pulse oximetry has also been an enormous boon in intensive care units. However, measurements can be difficult to obtain in low perfusion states or where inotropes such as dopamine are being used to sustain blood pressure.

Confounding factors (Table 2)

Abnormal haemoglobins and anaemia are 'physiological' confounders but these abnormalities can also affect the accuracy of the measurements. In animal experiments, SpO2 decreases as methaemoglobin increases up to 35%, and SpO2 increases as carboxyhaemoglobin increases up to 70%. Modest concentrations of these haemoglobins will not substantially change SpO2 which is a functional saturation. Anaemia has to be severe (50 g/L) before it interferes significantly with the measurement.

Abnormal dyes and pigments such as methylene blue (used to treat methaemoglobinaemia) and severe hyperbilirubinaemia may interfere. In most clinical circumstances, these disturbances will not be present to a significant degree, but they need to be kept in mind. Strong superficial pigments such as nail polish must be removed and signal failure may occur in black patients although careful positioning on the less pigmented nail bed usually overcomes this problem. Venous pulsation may confuse the signal, reducing the displayed saturation, particularly where a tourniquet is applied above the probe or in the presence of right heart failure or tricuspid incompetence. Excessive motion of the probe and strong incident light can also cause an erroneous or inadequate signal. Motion artifact is also a problem in many longer-term settings where movements can be interpreted as a pulse.

The pulse oximeter does not measure partial pressure of oxygen in arterial blood (usually expressed as mmHg) and the relationship between SpO2 and PaO2 is complex.

Clinical applications

Reliable pulse oximeters are now indispensable in all emergency departments, intensive care units (adult and neonatal) and operating theatres. Their use is considered to be good practice for procedures requiring sedation or instrumentation of the respiratory tract ranging from cardiac catheterisation to endoscopy and bronchoscopy. Their use in these procedures has uncovered quite alarming transient hypoxaemia requiring the use of supplemental oxygen. It is desirable to maintain the SpO2 above 90%.4

The routine use of pulse oximeters in operating theatres and recovery rooms has coincided with a dramatic decrease in perioperative morbidity and mortality although, interestingly, a cause and effect relation has not been confirmed.5Clearly, disastrous errors such as incorrect connections in anaesthetic machines can be quickly recognised.

Patients presenting to emergency departments with cardiorespiratory disorders are routinely monitored with pulse oximetry. However, it is important to identify when the additional information available from in vitro analysis of an arterial blood gas sample is critical for management. Patients with worsening asthma, deteriorating pneumonia or left heart failure and those with chronic obstructive pulmonary disease developing clouded consciousness on supplemental oxygen all need arterial carbon dioxide partial pressure (PaCO2), pH and base excess measurements (Fig. 2).

Pulse oximeters in sleep investigation laboratories have substantially contributed to the explosion of knowledge about sleep and breathing over the last few decades. They are also extensively used during exercise testing in pulmonary function and cardiac stress test units. Here, small falls in PaO2 in the higher range will be difficult to detect, but hazardous falls will be readily identified.

Finally, pulse oximeters have a place in the non-procedural doctor's office where the detection of acute or chronic hypoxaemia may be important - as in the assessment of patients requiring home oxygen therapy. An SpO2 above 90% in a patient with chronic obstructive pulmonary disease is reassuring whereas a lower measurement would suggest the need for confirmatory measurement of arterial blood gases. Experience with these devices, and their widespread adoption, have emphasised their status as a truly important advance in non-invasive patient monitoring and investigation.

Email: iveny@mail.med.usyd.edu.au

References

  1. International symposium on innovations and applications of pulse oximetry (ISIAPO 2002). Anesth Analg 2002;94 (Suppl 1S).
  2. Schnapp LM, Cohen NH. Pulse oximetry: uses and abuses. Chest 1990;98:1244-50.
  3. Severinghaus JW, Naifeh KH. Accuracy of response of six pulse oximeters to profound hypoxia. Anesthesiology 1987;67:551-8.
  4. Young IH, Le Souëf PN. Clinical oximetry. Med J Aust 1993;159:60-2.
  5. Cooper JB, Cullen DJ, Nemeskal R, Hoaglin DC, Gevirtz CC, Csete M, et al. Effects of information feedback and pulse oximetry on the incidence of anesthesia complications. Anesthesiology 1987;67:686-94.