Acknowledgments
We thank Drs. Corinne L. Leach and Sara Berkelhamer for their critical review of the chapter.
Gas exchange is a dynamic process that is dependent on complex interactions between the respiratory, cardiovascular, and central nervous systems. In a healthy individual breathing room air, the arterial blood levels of oxygen and carbon dioxide are maintained within a narrow normal range. In neonates receiving assisted ventilation or supplemental oxygen or suffering from impaired respiratory control, gas exchange may be compromised and close monitoring of blood gases becomes imperative to maintain homeostasis. Analysis of an arterial blood gas (ABG) sample is the gold standard for assessment of gas exchange. However, inherent risks associated with arterial access such as vasospasm, thrombosis, ischemia, and blood loss with resulting anemia necessitate alternate noninvasive methods to assess gas exchange. Blood gas analysis provides only a snapshot of a very dynamic process and results are frequently delayed (unless performed using a point-of-care device). The deleterious effects of hypoxia, hyperoxia, hypocarbia, and hypercarbia on morbidity, mortality, and neurodevelopmental outcomes have been well documented, and the importance of close frequent monitoring of these parameters cannot be overemphasized. This chapter briefly describes the techniques, indications, strengths, and limitations of the common noninvasive modalities used to monitor gas exchange ( Fig. 11-1 ) in the neonatal intensive care unit (NICU). These monitors provide a large amount of data that, with careful interpretation, can guide therapeutic strategy and achieve the best possible outcome.
Noninvasive Monitoring of Oxygenation
Pulse Oximetry
Oxygen has been widely used in the NICU since the mid-twentieth century. The misadventures with oxygen therapy and retinopathy of prematurity are well documented in the neonatal literature. There is controversy surrounding the target saturations during intensive care of preterm infants despite multiple large randomized controlled trials. In the premature infant, the amount and type of hemoglobin and the percentage of arterial oxygen saturation of hemoglobin (SaO 2 ) are the major determinants of tissue oxygen delivery ( Fig. 11-2 ). Clinically cyanosis is detectable by visual observation only when deoxyhemoglobin (DeoxyHb) is above 5 g/dL and is an unreliable assessment of oxygenation. Hence continuous monitoring of saturation by pulse oximetry (SpO 2 ) is crucial to provide optimal care.
Pulse oximetry measures the percentage of hemoglobin saturated with oxygen. It provides a transcutaneous, noninvasive estimate of SaO 2 and displays a plethysmographic waveform with a heart rate. The monitoring of hemoglobin saturation is possible by pulse oximetry because of the transparency of tissue to light in the near-infrared spectrum and the distinct absorption spectra of the chromophores such as oxyhemoglobin (HbO 2 ) and DeoxyHb ( Fig. 11-3 ). Pulse oximetry is based on the Beer-Lambert law, which states that absorption of light of a given wavelength is proportional to the concentration of the light-absorbing substance (chromophore) and the light path length. Pulse oximetry is based on two principles: (1) spectrophotometry—HbO 2 and reduced DeoxyHb have different absorption spectra at different wavelengths of light (red and near infrared)—and (2) photoplethysmography—the amount of light absorbed by blood in the tissue changes with the arterial pulse (see Fig. 11-3 ). The pulse oximeter probe consists of two light-emitting diodes and a photodetector that are positioned facing each other with the light passing intermittently at very high frequency through the interposed tissue. Saturation is estimated from the relative absorption of the two wavelengths during pulsatile flow (due to arterial blood—referred to as AC by convention) versus nonpulsatile flow (venous and tissue absorption—referred to as DC):
where f is the calibration constant.
The calibration algorithm for pulse oximetry is generated by subjecting healthy volunteers to varying inspired oxygen concentrations and correlating the arterial gas SaO 2 with the ratio of absorption (shown in the formula above) over a range of saturation values. As it is unethical to expose healthy volunteers to hazardously low saturations, readings below 75% are based on data extrapolated from calibration values obtained between 100% and 75%. Values obtained by pulse oximetry are within ±2% to 3% of the true SaO 2 value between 70% and 100%. Recently, “Blue” sensors (Masimo, Irvine, CA) calibrated for the 60% to 80% range are being used in neonates with cyanotic congenital heart disease (CHD) ( http://www.masimo.com/sensors/specialty.htm ). Harris et al reported that these sensors have better accuracy in the 75-85% SpO 2 range with 86% of the samples demonstrating <5% difference compared to co-oximetry (compared to only 69% of samples with <5% difference from co-oximetry values from standard pulse oximetry sensors). However, there was a further increase in differences for SaO 2 values <75% and neonates <3 kg were not tested.
Although pulse oximetry is now the standard of care in the NICU, there are no clearly established normal values in neonates. Pulse oximetry studies in normal healthy term infants and preterm infants breathing room air have shown average saturations to be 97% and 95%, respectively. However, defining a target range of SpO 2 in preterm infants on oxygen therapy or mechanical respiratory support as well as in term infants with persistent pulmonary hypertension of the newborn (PPHN) remains a topic of controversy.
Indications for Pulse Oximetry
In the NICU, pulse oximetry has become the standard of care and is considered the fifth vital sign. Estimation of SpO 2 by continuous pulse oximetry is used to (1) titrate inspired oxygen concentration in infants receiving supplemental oxygen, (2) define bronchopulmonary dysplasia (BPD) with the oxygen-reduction test, (3) monitor stable growing premature infants for bradycardia and desaturation spells, (4) titrate supplemental oxygen therapy during delivery room resuscitation and stabilization, (5) screen for critical CHD in the newborn period, (6) monitor an infant’s status during transport, (7) diagnose PPHN with ductal shunt by dual-pulse oximetry (postductal lower limb SpO 2 >5% to 10% lower than right upper limb value), and (8) perform car seat testing prior to discharge of at-risk infants.
Delivery Room Resuscitation
With the recognition of the dangers of oxidative stress associated with hyperoxia in the immediate newborn period, the use of oxygen blenders and pulse oximetry has been recommended for resuscitation and stabilization of newborn infants in the delivery room. Pulse oximetry assists in monitoring response to resuscitation (improvement in heart rate) and titration of supplemental oxygen therapy. The Neonatal Resuscitation Program (NRP) 2016 guidelines define target saturations based on time after birth to guide optimal use of oxygen in the delivery room. The use of pulse oximetry in the delivery room presents unique challenges. Delay in obtaining a stable tracing and readout of SpO 2 and heart rate is common. The average time to detect reliable signal on pulse oximetry has been estimated to be between 1 and 2 minutes. Using the correct order of connection minimizes time to obtaining a stable signal. (1) First, connect the oximeter cable to the pulse oximeter monitor; (2) turn on the pulse oximeter monitor; (3) apply the probe to the baby; and (4) connect the probe to the oximeter cable. The average difference in time with this optimized sequence has been shown to be 7 seconds as compared to connecting the sensor to the cable first and then applying the sensor to the infant. However, a more recent study performed in the delivery room with infants >/= 28 weeks gestation questioned this approach and suggested that applying pulse oximetry sensor to the oximeter first and then to the infant resulted in an earlier detection of a reliable signal. Preductal pulse oximetry should be recorded from the right upper limb as there is a substantial pre/postductal SpO 2 difference in the immediate newborn period. The NRP guidelines for delivery room SpO 2 were created using preductal SpO 2 . In addition, signal detection has been demonstrated to be faster with the pulse oximetry probe applied to the hand compared to the foot. Movement, poor perfusion, difficulties with probe placement, and high ambient light can all interfere with obtaining a reliable pulse oximetry signal.
Limitations of Pulse Oximetry
- 1.
Hyperoxia, hypoxia, and hypercarbia: Owing to the sigmoid shape of the oxyhemoglobin equilibration curve (see Fig. 11-3 ), pulse oximetry is unable to detect significant hyperoxia and is slow to detect acute hypoxemia. In infants receiving supplemental oxygen even large changes in PaO 2 result in little change in SpO 2 if the saturation is close to 100%. Alveolar hypoventilation may also be missed in infants on supplemental oxygen monitored solely with pulse oximetry and can lead to significant hypercarbia without an appreciable change in SpO 2 . Hence patients on supplemental oxygen at risk of hyperoxia/hypoxia/hypercarbia should have intermittent PO 2 and PCO 2 measured by blood gases. Anemia (unless very severe) and polycythemia do not affect pulse oximetry readings.
- 2.
Hypoperfusion and hypothermia: Pulse oximetry relies on normal pulsatile flow for its signal and hence can be falsely low in the setting of impaired perfusion or vasoconstriction associated with hypothermia, vasopressor treatment for hypotension, tourniquet effect from blood pressure cuff, etc. When applying the sensor circumferentially to the finger, hand, or foot, it should not be applied too tightly.
- 3.
Movement artifact: Conventional pulse oximetry is based on pulsatile flow of blood and calculates SpO 2 based on the assumption that arterial blood is the only component that moves at the site of measurement. During periods of body movement the blood in the venous and tissue compartment also moves and interferes with the SpO 2 reading or causes a signal dropout. This can disrupt monitoring during transport or during periods of spontaneous activity. Newer pulse oximeters with signal extraction technology use adaptive filtering to separate the components of data and filter noise from the signal to limit motion artifact.
Functional vs Fractional Saturation
Functional saturation refers to the percentage of hemoglobin that is saturated with oxygen in relation to the amount of hemoglobin that is capable of transporting oxygen—that is, HbO 2 /(HbO 2 + DeoxyHb). In contrast, fractional saturation is the percentage of oxygenated hemoglobin to the total hemoglobin, which includes variant hemoglobin molecules such as methemoglobin (MetHb) and carboxyhemoglobin (COHb) that are incapable of binding oxygen. Conventional oximeters display functional saturation and do not distinguish the variant hemoglobins from HbO 2 and provide saturation readings that are higher than the fractional saturation in patients with dyshemoglobinemias. High levels of COHb cause an increase in the SpO 2 approximately equal to the amount of COHb that is present. The presence of MetHb will bias the functional SpO 2 reading towards 85%, which will result in over- or underestimation of saturation for %HbO 2 values below and above 85% respectively. In normoxic subjects, high levels of MetHb decrease the SpO 2 reading by about half of the MetHb percentage concentration. New-generation oximeters are capable of detecting COHb and MetHb by using additional wavelengths of light and provide saturation readings that are more accurate in the clinical setting. When oxygen saturation determination by co-oximetry on a blood gas sample is discrepant by 5% or greater than the saturation measured by pulse oximetry, the presence of a variant hemoglobin has to be considered.
Additional Considerations
Fetal hemoglobin has light absorption characteristics similar to those of adult hemoglobin and does not affect pulse oximetry readings. However, one has to keep in mind its effect on the oxygen dissociation curve and tissue oxygenation at the displayed SpO 2 values (see Fig. 11-3 ). Indirect bilirubin has a different light absorption spectrum at 450 nm and typically will not affect SpO 2 readings. However, interference from ambient light with phototherapy and elevated COHb with hemolysis can alter SpO 2 in neonates. Discoloration from bronze-baby syndrome (due to phototherapy in the presence of direct hyperbilirubinemia) has also been reported to interfere with pulse oximetry readings.
Various indices have been derived from the data provided by oximetry and are of potential use in interpreting clinical status and guiding clinical care. These include the following:
The oxygen saturation index (OSI) : The oxygenation index (OI = FiO 2 × 100 × mean airway pressure [MAP in cm H 2 O] ÷ PaO 2 [in mm Hg]) has been used to monitor the severity of hypoxemic respiratory failure (HRF) and response to treatment. Because the calculation of OI requires PaO 2 obtained by an ABG, OSI has been suggested as an alternative. Rawat et al. have shown that in newborn infants with HRF, the OSI (FiO 2 × 100 × MAP ÷ preductal SpO 2 ) correlated closely with the OI. The relationship of OSI with OI in the saturation range of 70% to 99% is OI = 2 × OSI. The use of OSI will allow continuous assessment of the severity of HRF.
The perfusion index (PI) is a measure derived from pulse oximetry and compares the pulsatile to the nonpulsatile signal [(pulsatile signal (AC)/nonpulsatile signal (DC)) × 100] and gives an indication of the perfusion at the monitored site ( Fig. 11-4 ). The value of PI as an indicator of a patient’s circulatory status is being investigated for identification of CHD (left obstructive heart disease is not typically identified on routine pulse oximetry screening for critical congenital heart disease [CCHD]), subclinical chorioamnionitis, severity of illness, and intravascular volume status. The value can range from 0.02% (very weak pulse strength) to 20% (very strong pulse strength) and is influenced by stroke volume, vasoactive drugs, temperature, and vasoconstriction at the site of probe placement.
The plethysmographic variability index (PVI) is also derived from pulse oximetry. The arterial pulse volume changes during phases of the respiratory cycle, and this is more pronounced when the preload is inadequate (see Fig. 11-4 ). PVI measures the change in PI during a respiratory cycle and is expressed as a percentage as shown in the following equation: PVI = [(PI max − PI min )/PI max ] × 100%. Early studies suggest that PVI may prove helpful in assessing the hemodynamic significance of patent ductus arteriosus in preterm infants and intravascular volume status in neonatal patients. A high PVI in the presence of hypotension may be an indication for a fluid bolus to increase intravascular volume. Cannesson et al. showed that a PVI of >14% before volume expansion identified response to a fluid load in adults with a sensitivity of 81% and specificity of 100%. PVI can also predict fluid responsiveness in infants undergoing congenital heart surgery, with a threshold of 13% helping to discriminate between responders and nonresponders with a sensitivity of 84% and specificity of 64%.
Transcutaneous Oxygen Monitoring
Transcutaneous oxygen (TcP o 2 ) is occasionally used in NICUs as an alternative to arterial PaO 2 measurements in infants without arterial access but in need of continuous monitoring of PaO 2 . The sensor consists of a platinum cathode and a silver reference anode. The electrode is separated from the skin surface by a thin membrane through which the oxygen diffuses. The reduction of oxygen at the platinum sensor cathode generates a current that is processed to a PO 2 readout. Studies comparing PaO 2 with TcP o 2 in infants show good correlation. However, sensors need frequent repositioning and recalibration with ABGs, and the need for higher operating temperature is associated with a risk of thermal burns in preterm infants. With oxygenation being routinely monitored by pulse oximetry, transcutaneous monitoring of oxygen is becoming less common.
Noninvasive Assessment of PaCO 2
Arterial PCO 2 is a reflection of the interaction of CO 2 production in the body (metabolism), transport (systemic and pulmonary perfusion), and elimination (ventilation). Capnography provides instantaneous breath-to-breath analysis of exhaled CO 2 and has become an integral part of monitoring in the operating room. PaCO 2 in normal healthy infants ranges between 35 and 45 mm Hg. Cerebral blood flow is dependent on the arterial PaCO 2 and increases with hypercarbia and decreases with hypocarbia. In mechanically ventilated extremely preterm infants fluctuations of PaCO 2 are common and predispose the infants to intraventricular hemorrhage, periventricular leukomalacia, and BPD. Approaches to ventilation strategy (high frequency vs conventional) and adjustment of ventilator settings are frequently based on PaCO 2 . Assessment of PaCO 2 with intermittent ABGs can lead to unrecognized periods of hypercarbia and hypocarbia and missed opportunities for ventilator weaning. The continuous noninvasive assessment of PaCO 2 may be achieved using end-tidal CO 2 (Et co 2 ) or transcutaneous CO 2 (TcP co 2 ).
Capnography and End-Tidal CO 2 Monitoring
Capnography, the measurement of Et co 2 levels in exhaled breath, is based on the principle that the CO 2 diffuses easily from the pulmonary capillary into the alveolus and rapidly equilibrates with alveolar CO 2 (PA co 2 ). The capnogram is a graphic display of the levels of CO 2 during a respiratory cycle. During inspiration, PCO 2 on the capnogram is zero as the atmospheric air contains very little CO 2 . At the beginning of exhalation the gas from the anatomic dead space is expired first and has minimal CO 2 (phase I, see Fig. 11-1 ). In phase II, gas from the alveoli mixes with gas in the dead space, resulting in a sharp increase in the CO 2 concentration, and this reaches a peak and then plateaus as all of the expired gas is derived from the alveoli (phase III). The CO 2 level in the sampled gas is measured using infrared spectroscopy. CO 2 absorbs infrared light of a specific wavelength (4.26 μm), and this is used to calculate the amount of CO 2 in the sample ( Fig. 11-5 ).
