Blood gas measurements and complementary, noninvasive monitoring techniques provide the clinician with information essential to patient assessment, therapeutic decision making, and prognostication. Blood gas measurements are as important for ill newborns as for other critically ill patients, but rapidly changing physiology, difficult access to arterial and mixed venous sampling sites, and small blood volumes present unique challenges. This paper discusses considerations for interpretation of blood gases in the newborn period. Blood gas measurements and noninvasive estimations provide important information about oxygenation. The general goals of oxygen therapy in the neonate are to maintain adequate arterial PaO2 and SaO2, and to minimize cardiac work and the work of breathing. Pulse oximetry and transcutaneous oxygen monitoring are extraordinarily useful techniques of estimating and noninvasively monitoring the neonate’s oxygenation, but each method has limitations. Arterial blood gas determinations of pCO2 provide the most accurate determinations of the adequacy of alveolar ventilation, but capillary, transcutaneous, and end-tidal techniques are also useful. An approach to and examples of acid-base disorders are presented. Three hemoglobin variants relevant to the newborn are considered: fetal hemoglobin, carboxyhemoglobin, and methemoglobin. Blood gases obtained in the immediate perinatal period can help assess perinatal asphyxia, but particular attention must be paid to the sampling site, the time of life, and the possible and proven diagnoses.
- indexing terms: neonatal respiration
- bronchopulmonary dysplasia
- carbon dioxide monitoring
- oxygen monitoring
- acid-base disorders
- perinatal asphyxia
The perinatal period (labor, parturition, and the days following) is one of fundamental change in the cardiorespiratory status of the baby. Nutritional, excretory, and respiratory systems must rapidly assume new responsibilities as the organism changes from a dependent to a free-living individual. Respiratory gas exchange, formerly a placental function, must be established by the lungs within minutes after birth. The cardiovascular system undergoes changes just as dramatic, with conversion from two circulations in parallel to two circulations now in series. Therefore, frequent and serious difficulties in cardiorespiratory adaptation in the perinatal and neonatal periods are not surprising.
Blood gas measurements and complementary, noninvasive monitoring techniques provide the clinician with information essential to patient assessment, therapeutic decision making, and prognostication. Blood gas measurements are as important for ill newborn infants as for other critically ill patients, but unique challenges are provided by rapidly changing physiology, difficult access to arterial and mixed venous sampling sites, and small blood volumes. However, one must not negate the importance of historical and physical findings in the ill newborn. This information must be integrated with the laboratory data to best understand and treat the patient.
Normal values for arterial blood gases are very dependent on postnatal age (Fig. 1⇓ ). Values of Pao2 and Sao2 may also be lower in premature infants, caused by reduced lung function, and at high altitude, caused by reduced inspired oxygen tension. The most accurate method of measuring Pao2 and Sao2 involves placement of an indwelling catheter in either the aorta via an umbilical artery or in a peripheral artery; however, use of such catheters must be restricted to critically ill neonates because of frequent and serious thrombotic and infectious complications (1)(2). A problem associated with peripheral arterial catheters is hemodilution. For these catheters to remain patent, they are usually perfused with heparinized saline solution. Unless the catheter is cleared of perfusate, diluted samples will have lower Pco2 and bicarbonate values (3). Sampling methods should minimize blood loss and assure an undiluted arterial blood sample (4). Intermittent sampling of a peripheral artery often changes Pao2 significantly when the infant responds to pain by crying and can therefore underestimate or overestimate baseline Pao2 (5)(6). The site of arterial access must be considered if the ductus arteriosus, which connects the aorta and pulmonary artery, is still patent because a right-to-left shunt at this level will result in lower oxygen values in the descending aorta than in the blood perfusing the brain and eyes. In patients with chronic lung disease or mild-to-moderate acute cardiorespiratory problems, capillary blood gases are often utilized. Capillary values for pH and Pco2 are usually within 0.05 and 7.5 mmHg (1 kPa) of corresponding arterial values; however, Po2 underestimates Pao2 and, therefore, cannot exclude hyperoxemia (7). Capillary Po2 values are no longer useful, having been supplanted by the noninvasive techniques of transcutaneous (tc) Po2 and pulse oximetry monitoring that more reliably estimate Pao2 and Sao2, respectively.1 Pulse oximetry or tcPo2 monitoring should be combined with capillary blood gases to obtain an accurate and comprehensive evaluation of oxygenation. Capillary blood gases are not reliable for seriously ill patients, or for those with shock, hypotension, or peripheral vasoconstriction. In the first day of life, poor perfusion to the hands and feet (“acrocyanosis”) precludes the use of capillary blood gases. In these settings, arterial blood gases are required.
Precision, measured as the CV for replicate samples, of modern blood gas analyzers should be within 0.2% for pH, 4% for Pco2, and 3% for Po2 (Table 1⇓ ). Accuracy, measured as deviation from a known calibrator, for blood gas analyzers must be verified on a regular basis. Total analytic error for Pao2 and Paco2 approaches the clinically acceptable error (Table 1⇓ ).
assessment of oxygenation
Blood gas measurements and noninvasive estimations provide important information about oxygenation. Oxygen delivery (Do2) to tissues is the product of cardiac output (c.o.) and blood oxygen content (Cao2), Do2 = c.o. × Cao2. Ignoring the negligible oxygen dissolved in plasma, the equation can be expanded to Do2 = (HR × SV) × (Sao2 × 1.34 × Hgb), where HR = heart rate, SV = stroke volume, Sao2 = hemoglobin saturation, and Hgb = hemoglobin content. Insufficient oxygen delivery to tissues, hypoxia, can therefore be caused by cardiac failure (decreased HR and (or) SV leading to decreased c.o.), or by low hemoglobin (anemia) or low Sao2 (hypoxemia) leading to low Cao2 (Table 2⇓ ). When insufficient oxygen is provided to tissues, hypoxia leads to metabolic acidosis. Thus, blood gas measurements, specifically Po2, Sao2, pH, and base excess, can help to assess patient oxygenation but must be combined with other clinical and laboratory assessments to provide a comprehensive picture.
The general goals of oxygen therapy in the neonate are to maintain adequate Pao2 and Sao2, and to minimize cardiac work and the work of breathing (8). It is important to realize that “optimal oxygenation” will result in different Pao2/Sao2 goals for different types of neonatal patients. Most commonly, premature infants in respiratory failure should have Pao2 values between 6.66 and 10.66 kPa (50–80 mm Hg) (9). These goals minimize the chances of blindness caused by retinopathy of prematurity (10) and lower the inspired O2 and airway pressure requirements that, if higher, might increase the likelihood of bronchopulmonary dysplasia (BPD) (11). By contrast, full-term infants with diaphragmatic hernia or persistent pulmonary hypertension may require Pao2 values of 10.66–13.33 kPa (80–100 mm Hg) to maintain stability, minimize pulmonary resistance, and avoid worsening pulmonary hypertension (12). Infants with BPD or chronic lung disease show improved growth and less pulmonary hypertension (cor pulmonale) when Sao2 is kept >92% during wakefulness, sleep, and feeding (11). Liberal use of supplemental oxygen may be deleterious by promoting ductus arteriosus closure in some infants with congenital heart disease, such as hypoplastic left heart, by lowering pulmonary vascular resistance in other infants with large left-to-right shunts.
Pulse oximetry and transcutaneous oxygen monitoring are extraordinarily useful techniques of estimating and noninvasively monitoring the neonate’s oxygenation. In most settings they complement blood gases by permitting the clinician to noninvasively follow trends in patient oxygenation. However, neither technique can replace arterial blood gas monitoring in the critically ill patient because neither provides comprehensive and exact information on oxygenation, ventilation, acid–base status, and hemoglobin variants. Pulse oximetry has become more widely used because it usually reflects Sao2 accurately, is easy to use, and very rarely results in complications (Table 3⇓ ). Neither pulse oximetry nor transcutaneous oxygen monitoring is reliable for severe hypotension or peripheral vasoconstriction (13)(14). A false estimate of Sao2 can occur if the pulse oximeter probe is applied incorrectly, resulting in poor signal or an optical shunt, or if motion of the patient or probe occurs(14)(15). There has been concern that pulse oximetry monitoring, if not supplemented with intermittent arterial blood gas determinations, will not adequately protect the extremely premature infant from hyperoxia that predisposes to development of retinopathy of prematurity and blindness (10). For the smallest premature infants, whose retinas are still developing, exclusive reliance on noninvasive pulse oximetry to avoid hyperoxia is not recommended. Instead, keeping pulse oximetry Sao2 in the 88–92% range and intermittently using arterial blood gases to verify Sao2 and Pao2 is preferable (16)(17).
assessment of alveolar ventilation
Arterial blood gas determinations of Pco2 provide the most accurate determinations of the adequacy of alveolar ventilation. The Paco2 concentration in a given patient reflects the balance between metabolic production of CO2 and excretion by ventilation. Thus, a clinician might respond to an increased Paco2 by decreasing metabolic rate (sedation, paralysis, or reduction of thermal stress) or by increasing ventilation [increasing ventilator rate or tidal volume, decreasing added dead space, reducing airway resistance, or by surfactant administration in premature infants with respiratory distress syndrome, (RDS) to improve compliance].
The clinician must establish a target or acceptable range for Paco2 for a given patient. Although the normal range of Paco2 after the first hours of life can be considered 4.66–6 kPa (35–45 mm Hg), desirable CO2 values for a specific situation may be either higher or lower. For instance, in persistent pulmonary hypertension of the newborn, pulmonary artery pressures can be lowered by either respiratory or metabolic alkalosis (18). Modest respiratory alkalosis can rapidly lower pulmonary vascular resistance in some such patients. Because marked hypocapnea can decrease cerebral blood flow and has been associated with neurologic deficits, most clinicians no longer aim for Pco2 values <3.33 kPa (<25 mm Hg) (19). Infants with BPD (chronic lung disease) often tolerate Pco2 values of 6.66–8 kPa (50–60 mm Hg) (20), essentially “deciding” that normal blood gas status is not worth the markedly increased work of breathing necessary to achieve it. An approach termed “permissive hypercapnia” or “gentle ventilation” with lower ventilator pressures while tolerating slightly increased Paco2 resulted in decreased chronic lung disease for premature infants with RDS (21).
For most neonates and small infants, tcPco2 monitoring is usually preferred over end-tidal CO2 monitoring (PETco2) as a means of estimating and “trending” Paco2, and therefore alveolar ventilation (22)(23). Small tidal volumes, rapid respiratory rates, and inhomogeneous alveolar ventilation/perfusion in neonates with lung disease often preclude PETco2 monitoring in the newborn, especially in small prematures. By contrast, tcPco2 shows good correlation with Paco2 and provides an excellent trend monitor, accurately reflecting changes in Paco2. The tcPco2 monitor, unlike the tcPo2 monitor that must be heated to 43–44 °C, does not cause skin burns. When used at a temperature of 40–42 °C, the tcPco2 electrode can be left in place for 4 h in neonates and 8 h in infants and older children (24)(25). Because tcPco2 values sometimes are markedly inaccurate, in vivo calibration against an arterial or capillary blood gas is often required. Overestimation errors in hypercarbic patients are particularly frequent.
assessment of acid–base status
Blood gases provide essential information on acid–base status both in critically ill neonates and in chronically or less severely ill patients. One can approach the analysis of simple acid–base disorders by answering three questions. First, is the condition one of acidosis or alkalosis (is the pH less than or greater than 7.4)? Second, is the primary cause metabolic (is bicarbonate high or low) or respiratory (is Pco2 high or low)? Third, is the compensation appropriate? Fig. 2⇓ shows a clinically useful approach to blood gas interpretation in the newborn and infant (26).
To properly analyze and describe blood gases, certain terms must be defined. The suffix “emia” refers to the state of blood, for example, acidemia is a condition of excess blood acidity as indicated by pH. The suffix “osis” refers to a pathologic process in which acid or base is gained or lost from the body (27). Acidosis may not lead to acidemia, depending on the patient’s ability to compensate. Compensation is a response to the primary disorder, attempting to bring the pH as close as possible to neutral. Full compensation is often unachieved, and blood gases that appear to have fully compensated for the primary problem are most likely displaying a mixed picture, rather than complete correction.
Table 4⇓ presents some of the most common causes of acid–base disorders in neonates. Metabolic acidosis is most commonly caused by inadequate tissue perfusion (shock) caused by hypovolemia, decreased cardiac output, or sepsis. Hypoxemia caused by lung or heart disease often contributes to the tissue hypoxia and resulting lactic acidosis seen with hypoperfusion states. Sepsis in the newborn, as in older individuals, may cause metabolic acidosis by decreasing perfusion (“cold shock”) and by interfering with cellular aerobic metabolism (“warm shock”). To compensate for metabolic acidosis, term neonates and infants will attempt to lower Pco2 by hyperventilating; however, compensation is usually not complete, that is, not to a pH of 7.4. A suggested guideline for the desired Pco2 is as follows: The last two digits of the pH should equal the expected Pco2 (28). If the actual Pco2 is much higher than expected, there may also be a respiratory acidosis. Premature infants are often not able to compensate for a metabolic acidosis by hyperventilation and respiratory alkalosis. After treating the primary underlying problem causing the metabolic acidosis, slow infusions of sodium bicarbonate are often given.
One common problem in the management of infants with BPD is distinguishing a primary, chronic, respiratory acidosis with metabolic compensation from a diuretic-induced metabolic alkalosis with respiratory compensation. In infants with BPD, lung mechanics, ventilation/perfusion relations, and work of breathing are abnormal. This results in a chronically high Pco2—a primary respiratory acidosis. Renal compensation causes bicarbonate retention, bringing the pH back towards normal, but compensation is usually not complete, that is, pH remains <7.40. Diuretics are used to improve lung mechanics, to decrease lung water, and to improve gas exchange. Thiazide and especially loop diuretics result in a loss of chloride, potassium, and sodium, and in retention of bicarbonate. When high doses of diuretics are used without salt replacement, metabolic alkalosis can result, with pH values >7.40. Under these circumstances, respiratory drive can be depressed, worsening the hypoventilation. Lowering the dose of diuretics, changing from a loop to a thiazide diuretic, replacement of salt, or use of acetazolamide to lower plasma bicarbonate are strategies that can be used to minimize this problem.
hemoglobin variants in the newborn
Modern blood gas instruments often include options to measure hemoglobin and its variants such as fetal hemoglobin, carboxyhemoglobin, and methemoglobin. These capabilities can sometimes be used to our advantage in neonatal medicine. Fetal hemoglobin has a left-shifted oxyhemoglobin dissociation curve, with a 50% saturation point ∼2.8 kPa vs 3.47–3.6 kPa (21 mm Hg vs 26–27 mm Hg) for adult hemoglobin. Fetal hemoglobin is well designed to facilitate oxygen transport across the placenta. In the neonate, however, fetal hemoglobin releases less oxygen at any given capillary Po2. Pulse oximetry estimates of arterial hemoglobin saturation are as accurate for fetal as for adult hemoglobin. The Po2 to achieve “adequate” saturation will be lower for fetal than for adult hemoglobin. In practice, if pulse oximetry is being used to guide oxygen therapy, measurement of adult and fetal hemoglobin percentage adds little to clinical management. It should be realized, however, that Pao2 values in the 5.5–7 kPa (41–53 mm Hg) range are often high enough to achieve 88–92% Sao2 for premature infants with predominantly fetal hemoglobin.
In the neonatal setting, carboxyhemoglobin is of interest primarily in infants of smoking mothers. Carbon monoxide crosses the placenta and binds strongly to fetal hemoglobin, making it unavailable for oxygen transport (29). Effects of carbon monoxide include the functional anemia of carboxyhemoglobin, a left shift of the hemoglobin dissociation curve making oxygen less available to tissues, and an inhibition of mitochondrial cytochrome oxidase. Pulse oximeters use only two light wavelengths, thereby assuming that only deoxyhemoglobin and oxyhemoglobin are present. A CO-oximeter is required to measure carboxyhemoglobin.
The recent use of inhaled nitric oxide to treat pulmonary hypertension in the newborn and in older patients has refocused attention on methemoglobinemia. Inhaled nitric oxide binds to hemoglobin rapidly in the pulmonary circulation, resulting in selective relaxation of pulmonary vascular smooth muscle. The nitric oxide–hemoglobin complex is converted to methemoglobin, and toxic concentrations of nitric oxide can result in methemoglobinemia. To date, methemoglobinemia has not been a serious problem in neonates receiving 5–80 ppm inhaled nitric oxide. Intermittent CO-oximeter measurements of methemoglobin should be performed in patients receiving inhaled NO, especially at concentrations >40 ppm, to keep methemoglobin concentrations <5% of total hemoglobin (30).
blood gases in the immediate perinatal period with special reference to perinatal asphyxia
Blood gases can provide important information on patient status even before arterial blood sampling becomes possible after birth (Table 5⇓ ). Before the onset of labor, the fetus, compared with the normal adult, exists in a hypoxemic, normocarbic, nonacidotic environment. During the stress of normal labor, some tissue hypoxia and placental insufficiency occur, resulting in a mixed respiratory and metabolic acidosis. After birth, as pulmonary gas exchange is established, Pco2, pH, and Po2 values move toward normal adult values; the largest changes occur in the first few minutes after birth (Fig. 1⇑ ). Accordingly, the most important factors to consider when interpreting blood gases are the sampling site, the time of life, and the possible and proven patient diagnoses.
Perinatal asphyxia occurs when there is inadequate placental gas exchange to meet ongoing fetal tissue needs for oxygen consumption and carbon dioxide elimination. The combination of lactic acidosis, the product of anaerobic metabolism, and CO2 accumulation results in a mixed acidosis. It is important to note that current work suggests that only 10–20% of cerebral palsy cases is accounted for by perinatal asphyxia (31). Unfortunately, during labor, there is no noninvasive, simple method of monitoring fetal well-being that is both highly sensitive and highly specific. Fetal heart rate monitoring, either electronic or auscultatory, is reassuring when normal but has a false-positive rate >99%. Measurement of pH from capillary blood samples taken from the presenting part can provide additional information on fetal well-being when there is concern because of an abnormal fetal heart rate pattern. Values >7.24 are reassuring, whereas those <7.2 suggest that obstetric management options should be reevaluated (32).
In summary, although arterial blood gases can provide much useful information about the physiologic state of the patient, a clear and systematic approach is required to give meaning to the values. From procurement to analysis, potential sources of error must be considered and a complete understanding of what blood gases can and cannot tell you is needed to best treat the critically ill newborn.
We thank Pierre Senécal for helpful suggestions and Rosanna Barrafato for preparation of the manuscript.
The Department of Pediatrics, McGill University/Montreal Children’s Hospital, Montreal, QC, Canada.
↵1 Nonstandard abbreviations: c.o., cardiac output; SV, stroke volume; RDS, respiratory distress syndrome; tc, transcutaneous; and BPD, bronchopulmonary dysplasia.
1 Instrument precision for Pco2 and Po2 depends on the Pco2 and Po2 at which the test is made. Values given are representative of newer blood gas machines.
Adapted from Huch, 1994 (32).
- © 1997 The American Association for Clinical Chemistry