Pulse oximetry

Nov 23, 1999

Hypoxemia constitutes an important cause of morbidity and mortality in the intensive care unit (ICU), emergency department, and operating room. As a result, accurate detection is important. It has long been recognized that a patient's oxygenation is difficult to assess by physical examination alone. Frank cyanosis does not develop until the deoxyhemoglobin level is 5 g/dL, which corresponds to an arterial oxygen saturation (SaO2) around 67 percent [1]. Furthermore, the threshold at which cyanosis becomes apparent is affected by skin perfusion, skin pigmentation, and hemoglobin concentration [2].

Blood gas analysis was for many years the accepted method of detecting hypoxemia in critically ill patients, but this technique is painful, has potential complications, and does not provide immediate or continuous data [3]. (See "Measurement of arterial blood gases" and see "Placement and management of arterial catheters"). For these reasons, pulse oximetry has become the standard for continuous and/or noninvasive assessment of arterial oxygen saturation; it is now in such ubiquitous use that it has been called the "fifth vital sign" [4]. This card will review the theoretical and clinical aspects of pulse oximetry.

HISTORY — Carl Matthes invented the first noninvasive, nonpulsatile oximeter employing an ear probe in 1935. The device used two wavelengths of light to compensate for variations in tissue thickness and blood content. The development of oximetry intensified during World War II, when a method was sought to monitor oxygenation in pilots flying at high altitudes in pressurized cockpits [3,5]. These early nonpulsatile devices did not measure true arterial saturation because of interference from capillary and venous blood, and they were unwieldy to use and transport [6].

In 1970, scientists at Hewlett-Packard developed the first widely-used, commercial ear oximeter that preferentially measured arterial saturation by heating the tissue to 41ºC to increase blood flow [5]. In 1974, Takuo Aoyagi discovered that arterial oxygen saturation could be measured by looking for pulsations in the light signals coming through tissue. This eliminated the need for heating the tissue, and his device is the ancestor of the current generation of pulse oximeters [6].

PRINCIPLES OF OPERATION — The theoretical basis for pulse oximetry derives from the Beer-Lambert law. This states that the absorption of light of a given wavelength passing through a nonabsorbing solvent which contains an absorbing solute is proportional to the product of the solute concentration, the light path length, and an extinction coefficient. The Beer-Lambert law can readily be applied to co-oximeters, in which a sample of arterial blood can be placed within a cuvette and factors such as light path length and solute concentration can be controlled [7].

Application of this spectrophotometric principle to a noninvasive device, where tissue is of varying thickness and blood flows in a pulsatile manner, is more difficult. Modern pulse oximeters address these factors through the use of two wavelengths of light and complex microprocessors. Deoxyhemoglobin absorbs light maximally in the red band of the spectrum (600 to 750 nm), while oxyhemoglobin absorbs maximally in the infrared band (850 to 1000 nm) [8]. Absorbance at these two wavelengths is used to estimate saturation, which is derived from the ratio of oxyhemoglobin to the sum of oxyhemoglobin plus deoxyhemoglobin [1].

Pulse oximeter probes consist of a photodetector and two light-emitting diodes, one emitting at 660 nm and the other at 940 nm. The detector and emitters are positioned facing each other through interposed tissue [2]. Probes are most frequently placed on fingers or ear lobes [9]. In infants, probes may also be placed on the palms, toes, feet, arms, cheeks, tongue, penis, nose, or nasal septum [10].

The photodiodes are switched on and off several hundred times per second, so that light absorption by oxyhemoglobin and deoxyhemoglobin is recorded during pulsatile and nonpulsatile flow [11]. Absorption during pulsatile flow relates to the characteristics of arterial blood plus background tissue and venous blood, whereas absorption during nonpulsatile flow is due only to the background tissue and venous blood [8,12] (show figure 1). Absorption at the two wavelengths during pulsatile flow is divided by absorption during nonpulsatile flow, and these ratios are fed into an algorithm in the microprocessor to yield a saturation value. The displayed value is an average based on the previous three to six seconds [2,11]. In addition to SaO2, many pulse oximeters also display pulse rate and relative pulse amplitude [5].

The microprocessors of pulse oximeters are calibrated using reference tables compiled by exposing healthy volunteers to decreasing FiO2 to yield SaO2 ranging from 100 to 75 percent by co-oximetry. Because it would be unethical to intentionally generate lower saturations in volunteers, values for an SaO2 less than 75 percent are obtained by extrapolation from these volunteer data. Pulse oximeter manufacturers claim that reported values between 70 and 100 percent are accurate to within ± 2 percent of the true value, while those between 50 and 70 percent are within ± 3 percent. In practice, the cut-off for acceptable accuracy is felt by many clinicians to be 80 percent (which usually reflects a PaO2 of approximately 50 mmHg at a pH of 7.4), and varies depending on the model of pulse oximeter used [1,13,14]. The high accuracy of one type of pulse oximeter at SaO2 values ranging from 82 to 94 percent was confirmed in a study of 100 patients, although the the accuracy of the instrument deteriorated at values outside these parameters [15].

APPLICATIONS — Pulse oximetry is indicated in any clinical setting where hypoxemia may occur. These settings include patient monitoring in emergency departments, operating rooms, postoperative recovery areas, endoscopy suites, sleep and exercise laboratories, oral surgery, cardiac catheterization, conscious sedation, labor and delivery, and interfacility patient transfer [2,16]. As an example of its benefits, pulse oximetry in a pediatric ICU decreases the number of blood gases obtained and the duration of oxygen therapy without jeopardizing patient outcome [17].

In neonatal ICUs, oxygen toxicity in premature neonates can be limited by titrating oxygen to an SaO2 of 90 percent (which usually reflects of PaO2 of approximately 60 mmHg at a pH of 7.4) [6,10]. (See "Oxygen toxicity"). However, just as pulse oximeters are not ideally suited to detect severe hypoxemia, they are also not well suited to detect hyperoxemia. Due to the shape of the O2-hemoglobin dissociation curve, large changes in PaO2 may result in no change in O2 saturation if the saturation is already near 100 percent. Therefore, in patients at risk for profound hypoxemia or for hyperoxemia, SaO2 by pulse oximeter should be verified by arterial blood gas analysis [8].

Pulse oximetry offers the advantage of providing data on hemoglobin saturation rather then PO2. SaO2 reflects the 98 percent of arterial oxygen content that is normally carried by hemoglobin, while the PO2 directly measures only the small amount of oxygen that is dissolved in plasma. PO2 commonly is used to estimate arterial oxyhemoglobin saturation and oxygen content because the dissolved and hemoglobin-bound oxygen pools are in equilibrium, but changes in pH, temperature, and 2,3-diphosphoglycerate concentration alter the PO2-SaO2 relationship and may result in misleading calculations of oxyhemoglobin saturation.

LIMITATIONS — While pulse oximetry is a convenient way of measuring arterial oxygenation, it does not assess ventilation. Therefore, its use should be supplemented with arterial blood gas analysis when hypoventilation is a concern [16]. As an example, one inadvertently hypoventilated patient who was administered 100 percent oxygen during hip arthroplasty and monitored with pulse oximetry alone developed a PaCO2 of 265 mmHg and an arterial pH of 6.65 despite maintenance of oxygen saturations of 94 to 96 percent [18].

In addition, because it does not measure PaO2, overreliance on pulse oximetry may delay detection of clinically significant hypoxemia. A large decrease in PO2 will not produce a significant fall in SaO2 until the steeper portion of the oxygen hemoglobin dissociation curve is encountered at a PO2 of approximately 60 to 70 mmHg. This is particularly important in patients receiving supplemental oxygen. As an example, a fall in PaO2 in such a patient from 140 to 65 mmHg would be required before a significant decrease in oxygen saturation is detected.

Pulse oximetry results are signal-averaged over several seconds. Therefore, the pulse oximeter may not detect a hypoxemic event until well after it has occurred [6]. This delay may be of particular significance when the device is being used for monitoring during intubation.

TECHNICAL SOURCES OF ARTIFACT — Interpretation of pulse oximetry readings must account for a variety of factors which may artifactually influence the results. The best defense against these potential sources of error is a high index of suspicion. If a saturation reading is in doubt, a quick quality assurance test can be done by the health care worker putting the probe on his or her own finger as near as possible to the original patient site [12]. The following are some of the more common potential technical sources of error.

Ambient light — Intense daylight, fluorescent, incandescent, xenon, and infrared light sources have resulted in spurious pulse oximetry readings [2]. In such cases, the oximeter will often give a falsely low reading of 85 percent, the saturation at which the ratio of red to infrared light is 1. Falsely elevated readings due to ambient light of normal intensity have been reported, but are rare [8].

Improper probe placement — Due to partial detachment of the probe, light from only one of the two light-emitting diodes may pass through the tissue, causing either a falsely elevated or depressed reading [8]. A similar problem may occur in infants and small children, because the small size of fingers or other tissues may result in differences in the path length of one light source compared to that of the other. These problems can be minimized by ensuring that the probe is properly attached with the light sources and detectors opposite each other in a nontangential path [19].

Placement of the sensor on the same extremity as a blood pressure cuff or arterial line can cause erroneous readings and should be avoided [20]. The choice of probe site may also affect accuracy; finger probes appear more accurate than forehead, nose, or earlobe probes during low perfusion states [9].

Motion artifact — A poor signal-to-noise ratio will cause signal artifact [1,5]. This most commonly results from motion due to shivering, seizure activity, pressure on the sensor, or transport of the patient by ambulance or helicopter.

Electromagnetic radiation — Radio frequency emissions from magnetic resonance imaging (MRI) scanners may interfere with pulse oximetry. In addition, second- and third-degree burns beneath pulse oximeter probes have been reported in patients undergoing MRI studies [10]. This complication is believed to result from the generation of electrical skin currents beneath the looped pulse oximeter cables which act as an antenna.

Other sources of electromagnetic radiation, such as cellular phones and electrocautery devices, can also interfere with pulse oximeters [12,19].

PATIENT-RELATED SOURCES OF ERROR — Pulse oximetry readings may be inaccurate in certain clinical situations even if the device is functioning properly and is free from external interference. The results of pulse oximetry should be interpreted cautiously in the presence of abnormal hemoglobins, nail polish, deeply pigmented skin, hypoperfusion, anemia, venous congestion, or when certain dyes are in use.

Abnormal hemoglobins — Abnormal hemoglobins or hemoglobin variants may interfere with pulse oximetry if their absorption properties are similar to those of oxyhemoglobin or deoxyhemoglobin.

  •  Carboxyhemoglobin absorbs approximately the same amount of 660 nm light as does oxyhemoglobin; the pulse oximetry reading therefore represents an inexact summation of oxyhemoglobin and carboxyhemoglobin [1,8,9,21,22] (show figure 2). In cases of carbon monoxide poisoning or in chronic, heavy smokers, a falsely reassuring pulse oximetry reading may mask life-threatening arterial desaturation.

When carboxyhemoglobinemia is suspected, co-oximetry is required to accurately measure oxyhemoglobin. Co-oximeters, which use four rather than two wavelengths of light, detect oxyhemoglobin, deoxyhemoglobin, carboxyhemoglobin, and methemoglobin, but require a sample of arterial whole blood [6,17]. (See "Smoke inhalation").

  •  Methemoglobin absorbs at both 660 and 940 nm [9]. Up to a methemoglobin level of 20 percent, SaO2 drops by about one-half of the methemoglobin percentage. At higher methemoglobin levels, SaO2 tends toward 85 percent regardless of the true percentage of oxyhemoglobin [1,10,23,24] (show figure 3).

  •  Sickle cell hemoglobin generally produces pulse oximeter readings similar to normal hemoglobin, but cases of falsely elevated and falsely low readings have been reported [16,25].

  •  Fetal hemoglobin gives pulse oximetry readings clinically indistinguishable from those of adult hemoglobin [5].

Nail polish — The use of nail polish can potentially affect pulse oximeter readings if the polish absorbs light at 660 nm or 940 nm [1]. A small study of volunteers wearing black, green, and blue nail polish revealed a drop in SaO2 of 3 percent, 5 percent, and 6 percent, respectively [12]. This problem can be avoided by mounting the probe on the finger sideways, rather than in a dorsal-ventral orientation [10]. Red nail polish does not appear to have an effect on pulse oximetry readings.

Skin pigmentation — Data on the effect of skin pigmentation on pulse oximetry are conflicting [12]. In theory, skin pigmentation should have no effect, since it should absorb at a constant level and be subtracted out as part of the background in the SaO2 calculation; this hypothesis appears to apply to altered pigmentation due to hyperbilirubinemia [8]. However, readings which are erroneously elevated by approximately 4 percent and a higher incidence of signal detection errors have been described in African-American patients [10].

Hypoperfusion — Pulse oximetry readings can be falsely low due to signal failure in the setting of hemodynamic instability or poor limb perfusion from extremity elevation or vasoconstriction. Measures which may improve the signal in these settings include vigorous rubbing of the affected extremity, application of heat, administration of a digital block to the finger being used, or the use of topical vasodilators such as nitroglycerin paste or oil of wintergreen [2,3,16].

Anemia — In vitro and animal studies suggest that pulse oximetry readings may be affected by hemoglobin concentration [16]. In vivo, low hemoglobin concentrations appears to cause falsely low readings when the SaO2 is less than 80 percent [10]. However, this effect is not clinically significant until the hemoglobin level is less than 5 g/dL [26].

Venous congestion — Venous congestion due to tricuspid valve incompetence or cardiomyopathy may yield falsely low SaO2 readings due to generation of venous pulsations. This results from the instrument treating less oxygenated, pulsatile venous blood as part of the arterial sample, thereby producing a smaller calculated SaO2 [10].

Vital dyes — The vital dyes methylene blue, indocyanine green, fluorescein, and indigo carmine can cause erroneously low pulse oximetry readings. Methylene blue has the greatest impact, as it absorbs significantly at 670 nm. However, these effects tend to be transient and resolve as the dyes are diluted and metabolized [5,12].