Accuracy and Precision

Because pulse oximeters themselves cannot be calibrated, their accuracy is highly variable and dependent on both the calibration curve programmed into the monitor and the quality of signal processing.^5,6 The ratio of absorbencies is calibrated empirically against SaO₂ measured by co-oximetry in normal volunteers subjected to various levels of oxygenation. Pulse oximeters are calibrated against measured SaO₂ down to 70% (saturations below this level are determined by extrapolation).⁵ The resulting calibration curve is stored in the monitor’s microprocessor to calculate SpO₂.⁶

The accuracy of the calibration curve depends on laboratory testing conditions (co-oximeter used, range of oxygenation studied, and characteristics of sample subjects). Most manufacturers report an accuracy of ±2% at an SaO₂ greater than 70% and ±3% when the SaO₂ is 50% to 70%.² In normal subjects tested at an SaO₂ between 99% and 83%, pulse oximetry has a bias and precision that are within 3% of co-oximetry.⁷ However, under hypoxic conditions (SaO₂ 78% to 55%), when the monitor must rely on extrapolated values, bias increases (8%) and precision deteriorates (5%).⁷ Likewise, in critically ill patients, pulse oximeters historically perform well when the SaO₂ is greater than 90% (bias of 1.7%; precision of ±1.2%), but accuracy diminishes at an SaO₂ below 90% (bias of 5.1%; precision of ±2.7%)⁸ (Figure 43-3). Technologic advances over the past decade have apparently improved this performance; a recent study comparing pulse oximetry to co-oximetry reported a bias of 0.19% and a precision of ±2.22% over an SaO₂ range of 60% to 100%.⁹

Figure 43-3 Oxyhemoglobin dissociation curve relates oxygen saturation and partial pressure of oxygen in the blood, and is affected by many variables.

(Courtesy Phillips Medical Systems, Carlsbad, California.)

Dynamic Response

Because pulse oximeters detect very small optical signals (and must reject a variety of artifacts), data must be averaged over several seconds, thus affecting response time.⁵ Pulse oximeters may register a near-normal SpO₂ when the actual SaO₂ is less than 70%.⁵ A prolonged lag time is more common with finger probes than ear probes^5,10,11 and is attributed to hypoxia-related peripheral vasoconstriction.⁵ Bradycardia also is associated with a prolonged response time.¹¹

Sources of Error

Motion artifact and low perfusion are the most common sources of SpO₂ inaccuracies, because the photoplethysmographic pulse signal is very low in these settings compared with the total absorption signal.^12,13 The combination of motion artifact and low perfusion substantially lowers SpO₂ accuracy compared with either artifact alone.¹⁴ Causes of motion artifact include shivering, twitching, agitation, intraaortic balloon pump assistance, and patient transport.^15,16 Signs of motion artifact include a false or erratic pulse rate reading or an abnormal plethysmographic waveform. Peripheral hypoperfusion from hypothermia, low cardiac output, or vasoconstrictive drugs may increase bias, reduce precision, and prolong the detection time for a hypoxic event.¹⁶ Newer technologies have helped reduce the incidence of these problems but they have not been eliminated as a source of error. Relocation of the probe may be required to obtain a more accurate signal.

Despite recent technologic advances, there still are a number of factors that may affect the accuracy of the pulse oximeter. Table 43-1 lists the most common factors.

TABLE 43-1 Common Factors Affecting Pulse Oximetry Measurements

Reflectance Pulse Oximetry

Reflectance pulse oximetry was designed to counter signal-detection problems associated with finger probes during hypoperfusion. The reflectance sensor is designed for placement on the forehead just above the orbital area, where superficial blood flow is abundant and less susceptible to vasconstriction.²⁶ Whereas traditional probes work by transilluminating a tissue bed and measuring the forward-scattered light on the opposite side of the finger or earlobe, reflectance probes are constructed with the light-emitting diodes and the photodetector located on the same side. The photodetector measures the back-scattered light from the skin.²⁶ In addition, more liberal placement sites for reflectance pulse oximetry has allowed fetal monitoring during labor.²⁷ Intraesophageal SpO₂ monitoring is currently under investigation.²⁸ Anasarca, excessive head movement, and difficulty in securing the probe site are some of the problems encountered with reflectance pulse oximetry.²⁹ Light “shunting” from poor skin contact and direct sensor placement over a superficial artery are associated with artifacts.³⁰ Reflectance pulse oximetry is also limited by poor signal-to-noise ratio and variability among sites in the arrangement of blood vessels and tissue blood volume.³⁰ However, recent studies have shown reflectance pulse oximetry to be as effective as finger sensors in many situations.^31–34

Technologic Advances

Recent advances in signal analysis and processing have markedly improved SpO₂ accuracy during low perfusion and reduced the problem of motion artifact.^16,35 According to recent independent testing, these advances occur with pulse oximeters made by several manufacturers.³⁶ Durban and Rostow reported that new pulse oximeter technology can accurately detect SaO₂ in 92% of the cases in which traditional SpO₂ monitoring failed owing to low perfusion and motion artifact³⁷ (Box 43-1).

Box 43-1

Aarc Clinical Practice Guideline: Pulse Oximetry

Indications

• The need to monitor the adequacy of arterial oxyhemoglobin saturation.

• The need to quantitate the response of arterial oxyhemoglobin saturation to therapeutic intervention or to a diagnostic procedure (e.g., bronchoscopy).

• The need to comply with mandated regulations or recommendations by authoritative groups.

Contraindications

• The presence of an ongoing need for measurement of pH, PaCO₂, total hemoglobin, and abnormal hemoglobins may be a relative contraindication to pulse oximetry.

Precautions

• Pulse oximetry is considered a safe procedure, but because of device limitations, false-negative results for hypoxemia and/or false-positive results for normoxemia or hyperoxemia may lead to inappropriate treatment of the patient.

• Factors that may affect pulse oximeter readings include motion artifact, abnormal hemoglobins and methemoglobin, intravascular dyes, exposure of measuring probe to ambient light during measurement, low perfusion state, skin pigmentation, and nail polish or nail coverings with finger probe.

Assessment of Need

• When direct measurement of SaO₂ is not available or accessible in a timely fashion, an SpO₂ measurement may temporarily suffice if the limitations of the data are appreciated.

• SpO₂ is appropriate for continuous and prolonged monitoring (e.g., during sleep, exercise, bronchoscopy).

• SpO₂ may be adequate when assessment of acid-base status and/or PaO₂ is not required.

Assessment of Outcome

• The following should be utilized to evaluate the benefit of pulse oximetry:

SpO₂ results should reflect the patient’s clinical condition (i.e., validate the basis for ordering the test).

Documentation of results, therapeutic intervention (or lack of), and/or clinical decisions based on the SpO₂ measurement should be noted in the medical record.

From AARC clinical practice guideline: pulse oximetry. Respir Care 1992;37:891-7.

Monitoring

• The monitoring schedule of patient and equipment during continuous oximetry should be correlated with bedside assessment and vital signs determinations.

Clinical Applications

Capnometric determination of the partial pressure of CO₂ in end-tidal exhaled gas (PETCO₂) is used as a surrogate for the partial pressure of CO₂ in arterial blood (PaCO₂) during mechanical ventilation^40,41 (Figure 43-4). Although widely available today, the utilization of PETCO₂ to represent PaCO₂ in ICUs remains unclear. While perhaps not an exact match for PaCO₂, PETCO₂ does provide a valuable trending tool. Also, with newer technologies, the accuracy of PETCO₂ measurements is improving. In a recent study, McSwain et al. showed strong correlations between PETCO₂ and PaCO₂ across a wide range of dead-space conditions.⁴² Capnometry is used for a variety of purposes, such as the diagnosis of pulmonary embolism, determination of lung recruitment response to positive end-expiratory pressure (PEEP), detection of intrinsic PEEP, evaluation of weaning, indirect marker of elevated dead-space ventilation, assessment of cardiopulmonary resuscitation, indirect determination of cardiac output through partial CO₂ rebreathing, verification of endotracheal cannulation, detection of airway accidents, and even determination of feeding tube placement.^43–55 Guidelines for the use of capnometry/capnography are outlined by the American Association for Respiratory Care (Box 43-2).

Figure 43-4 Single-breath carbon dioxide waveform depicts carbon dioxide elimination as a function of the volume of gas exhaled. Phase 1 represents gas exhaled from upper airways. Phase 2 is the transitional phase from upper to lower airway ventilation and tends to depict changes in perfusion. Phase 3 is the area of alveolar gas exchange and represents changes in gas distribution. PETCO₂, partial pressure of end-tidal carbon dioxide.

Box 43-2

Aarc Clinical Practice Guideline: Capnography/Capnometry During Mechanical Ventilation

Indications

• Capnography should not be mandated for all patients receiving mechanical ventilatory support, but it may be indicated for:

Evaluation of the exhaled CO₂, especially end-tidal CO₂ (PETCO₂).

Monitoring severity of pulmonary disease and evaluating response to therapy, especially therapy intended to improve the ratio of dead space to tidal volume (VD/VT) and the matching of ventilation to perfusion (V/Q) and, possibly, to increase coronary blood flow.

Use as an adjunct to determine that tracheal rather than esophageal intubation has taken place.

Continued monitoring of the integrity of the ventilatory circuit, including the artificial airway.

Evaluation of the efficiency of mechanical ventilatory support by determination of the difference between PaCO₂ and PETCO₂.

Monitoring adequacy of pulmonary, systemic, and coronary blood flow.

Estimation of effective (nonshunted) pulmonary capillary blood flow by a partial rebreathing method.

Use as an adjunctive tool to screen for pulmonary embolism.

Monitoring inspired CO₂ when CO₂ gas is being therapeutically administered.

Graphic evaluation of the ventilator-patient interface.

Measurement of the volume of CO₂ elimination to assess metabolic rate and/or alveolar ventilation.

From AARC clinical practice guideline: capnography/capnometry during mechanical ventilation. Respir Care 2003;48:534-9.

Contraindications

• There are no absolute contraindications to capnography in mechanically ventilated patients, provided the data obtained are evaluated with consideration given to the patient’s clinical condition.

Precautions and Possible Complications

• With mainstream analyzers, the use of too large a sampling window may introduce an excessive amount of dead space into the ventilator circuit.

• Care must be taken to minimize the amount of additional weight placed on the artificial airway by the addition of the sampling window or, in the case of a sidestream analyzer, the sampling line.

Assessment of Need

• Capnography is considered a standard of care during anesthesia. The American Society of Anesthesiologists has suggested that capnography be available for patients with acute ventilatory failure on mechanical ventilatory support. The American College of Emergency Physicians recommends capnography as an adjunctive method to ensure proper endotracheal tube position.

• Assessment of the need to use capnography with a specific patient should be guided by the clinical situation. The patient’s primary cause of respiratory failure and the acuteness of his or her condition should be considered.

Assessment of Outcome

• Results should reflect the patient’s condition and should validate the basis for ordering the monitoring.

• Documentation of results (along with all ventilatory and hemodynamic variables available), therapeutic interventions, and/or clinical decisions made based on the capnogram should be included in the patient’s chart.

Monitoring

• During capnography, the following should be considered and monitored:

Ventilatory variables: tidal volume, respiratory rate, positive end-expiratory pressure, inspiratory-to-expiratory time ratio (I : E), peak airway pressure, and concentrations of respiratory gas mixture.

Hemodynamic variables: systemic and pulmonary blood pressures, cardiac output, shunt, and ventilation-perfusion imbalances.

PaCO₂-PETCO₂ Gradient

Normal subjects have a PaCO₂-PETCO₂ gradient of 4 to 5 mm Hg.^{40,43,47,56–60} In critically ill patients, the PaCO₂-PETCO₂ gradient can be markedly elevated, with a tendency toward wider gradients in obstructive lung diseases (7-16 mm Hg) than in acute lung injury or cardiogenic pulmonary edema (4-12 mm Hg).^{46,47,61–63} A strong correlation between ΔPETCO₂ and ΔPaCO₂ (r = 0.82), along with minor bias and reasonable precision between PETCO₂ and PaCO₂, suggests that arterial blood gas monitoring may not be needed to assess ventilation unless the ΔPETCO₂ exceeds 5 mm Hg.⁴⁸ Yet several studies found that the ΔPETCO₂ often falsely predicts the degree and direction of ΔPaCO₂.^58–6063 Therefore, despite PETCO₂ monitoring, routine arterial blood gas analysis is still required in critically ill patients.

Several factors determine the PaCO₂-PETCO₂ gradient. Whereas PaCO₂ reflects the mean partial pressure of CO₂ in alveolar gas (PaCO₂), PETCO₂ approximates the peak PaCO₂.⁶⁴ During expiration, lung regions with high ventilation-to-perfusion ratios dilute the mixed CO₂ concentration so that PETCO₂ is usually lower than PaCO₂.⁶⁵ However, when CO₂ production is elevated (or expiration is prolonged), PETCO₂ more closely resembles mixed venous PCO₂, as a higher amount of CO₂ diffuses into a progressively smaller lung volume.⁶⁴ Thus, the PaCO₂-PETCO₂ gradient can be affected by changes in respiratory rate and tidal volume (VT) due to alterations in expiratory time and by CO₂ production and mixed venous CO₂ content.⁶⁴ In fact, it is not uncommon for PETCO₂ to exceed PaCO₂.⁶⁵ Inotropic or vasoactive drugs may affect the PaCO₂-PETCO₂ gradient in an unpredictable manner, either by increasing cardiac output and pulmonary perfusion (thereby reducing alveolar dead space) or by reducing pulmonary vascular resistance and magnifying intrapulmonary shunt by countering hypoxic pulmonary vasoconstriction.⁵⁸

Mechanical factors can cause either inconsistencies or inaccuracies in PETCO₂. The sample tubing length and aspirating flow rates used in sidestream capnometers affect the time required to measure changes in tidal CO₂ concentration.⁶⁶ At respiratory frequencies above 30, capnometers tend to underreport the true PETCO₂.⁶⁷ This may occur because of gas mixing between adjacent breaths during transport down the sampling line and in the analysis chamber.⁶⁷ This problem can be avoided with mainstream analyzers, which provide near-instantaneous CO₂ measurement (<250 msec).⁶⁸

PaCO₂-PETCO₂ Gradient, Positive End-Expiratory Pressure, and Lung Recruitment

PEEP recruits collapsed alveoli, improves ventilation-perfusion matching, and reduces alveolar dead space, although excessive levels cause overdistention and increased alveolar dead space.⁶⁹ Because the PaCO₂-PETCO₂ gradient correlates strongly with the physiologic dead space–to–tidal volume ratio (VD/VT