Respiratory System

The small airways predispose the micropremie to obstruction and difficulty with ventilation. Resistance to airflow is inversely proportional to the fifth power of the radius in the upper airway and to the fourth power of the radius beyond the fifth bronchial division (see also Fig. 12-7). As a result, insertion of an endotracheal tube (ETT) increases resistance and work of breathing far greater for the micropremie (2.5 or 3 mm inside diameter [ID]) than for a larger infant (4 mm ID), child (5 mm ID), and adult (7 mm ID) (Fig. 35-1). Similarly, partial occlusion of the ETT by secretions, blood, or kinking increases the work of breathing to a much greater extent in the micropremie. Partial occlusion of the natural airway from loss of muscle tone during anesthesia and sedation also increases the work of breathing more in the micropremie. Consequently, general anesthesia often requires placement of an ETT to ensure airway patency, and assisted ventilation to overcome the increased work of breathing.

FIGURE 35-1 Change in work of breathing after placement of an appropriate size endotracheal tube in extremely low–birth-weight infants (less than 1000 g), premature infants (1500 g), full-term infants, children, and adults (see text for details).

(Redrawn with permission from Spaeth JP, O’Hara IB, Kurth CD. Anesthesia for the micropremie. Semin Perinatol 1998;22:390-401.)

Diseases that narrow the airway, such as subglottic stenosis, tracheal stenosis, and tracheobronchomalacia, occur commonly in the micropremie, and the associated reduction in airway diameter further increases both resistance to airflow and work of breathing. Subglottic stenosis necessitates the placement of a smaller ETT than would otherwise be placed, further increasing airflow resistance. Tracheal stenosis often occurs near the carina and, although not necessitating a smaller ETT, it increases airway resistance from the stenosis distal to the ETT. With tracheobronchomalacia, the intrathoracic airways collapse during exhalation, again increasing resistance and the work of breathing (see also Fig. 12-10). Positive end-expiratory pressure (PEEP) or continuous positive airway pressure (CPAP) helps stent open the airway. Mechanical ventilation, rather than spontaneous ventilation during anesthesia, prevents fatigue from increased work of breathing, and maintains ventilation and oxygenation. During anesthesia the use of smaller inspiratory-to-expiratory ratios prevent air trapping and hyperinflation of lung segments.

The structure and function of the immature lung predisposes to alveolar collapse and hypoxia. The alveoli are primarily composed of thick-walled, fluid-filled saccular spaces that are surfactant deficient and require greater pressures to initially expand. Production of surfactant begins at between 23 to 24 weeks gestation, although surfactant concentrations often remain inadequate until 36 weeks gestation. These factors lead to reduced lung volumes and lung compliance, increased intrapulmonary shunting, and ventilation-perfusion mismatch. Decreased lung volumes and ventilation-perfusion mismatch may also occur as a consequence of anesthesia. The effects of immature structure, disease, and anesthesia on lung function all increase the risk of hypoxia during surgery and anesthesia.

Micropremie lungs are particularly susceptible to oxygen toxicity, volutrauma, and the development of bronchopulmonary dysplasia (BPD). Mechanical lung injury is no longer thought to be caused by the use of high peak-inspiratory pressures, but rather related to increased end-inspiratory lung volumes and frequent collapse and reopening of alveoli. A ventilation strategy using small tidal volumes (4 to 6 mL/kg), greater respiratory rates, PEEP sufficient to avoid alveolar collapse, and permissive hypercapnia reduces lung injury in the premature lung.⁴ Randomized controlled trials of permissive hypercapnia (Paco₂ 45 to 55 mm Hg) showed smaller periods of assisted ventilation, reduced incidence of BPD, and no increase in adverse neurodevelopmental effects.⁵ The use of high inspired oxygen concentrations leads to the development of free radical species, which contribute to pulmonary epithelial cell injury.

A severity index for BPD based on the need for supplemental oxygen and/or positive-pressure ventilation or nasal CPAP has been developed and shown to identify a spectrum of risk for adverse pulmonary and neurodevelopmental outcomes in preterm infants (Table 35-1).⁶ Although this severity index has not been studied in the context of anesthetic risk, experience suggests that such infants requiring supplemental oxygen, positive pressure, or medications for reactive airways are at greater risk for perioperative pulmonary complications. Anesthetic goals include minimizing the inspired oxygen concentration and tidal volumes while maintaining oxygen saturation (Sao₂ 90% to 94%) and ventilation (Paco₂ 50 to 55 mm Hg). The use of smaller tidal volumes decreases the risk of pneumothoraces and interstitial emphysema.

TABLE 35-1 Severity-Based Diagnostic Criteria for Bronchopulmonary Dysplasia (BPD)

From Ehrenkranz RA, Walsh MC, Vohr BR, et al. Validation of the National Institutes of Health consensus definition of bronchopulmonary dysplasia. Pediatrics 2005;116:1353-60.

Respiratory Control

Micropremies possess a biphasic ventilatory response to hypoxia. Initially, ventilation increases in response to hypoxia, but after several minutes, ventilation decreases and apnea may ensue.⁷ The ventilatory response to carbon dioxide is decreased in the micropremie, and hypoxia further blunts this response.^8,9 Anesthetic drugs depress the ventilatory responses to both hypoxia and hypercapnia. Hypoxia and hypercapnia occur commonly as a result of apnea and hypoventilation during emergence and recovery from anesthesia. Thus the combination of anesthetic effects and an immature respiratory control system, as well as immature intercostal and diaphragmatic muscles,^10,11 increase the risk of hypoxia, hypercapnia, and apnea in the postoperative period.

Apneic episodes occur commonly in the micropremie but decrease with advancing postconceptional age.¹² These apneic episodes usually involve both a failure to breathe (central apnea) and a failure to maintain a patent airway (obstructive apnea). Central apnea results from decreased respiratory center output, although it may be precipitated by abrupt changes in oxygenation, pulmonary mechanics, brain hemorrhage, hypothermia, or airway stimulation. Apnea may also occur without a precipitating event (i.e., idiopathic). Preterm infants with apnea do not increase ventilation in response to hypercapnia, compared with those without apnea, thereby delaying resumption of breathing and prolonging the apneic episode.¹³ During obstructive apnea, the airway becomes obstructed in the hypopharynx and larynx as a result of pharyngeal muscle incoordination. Anesthetic drugs may further decrease pharyngeal muscle tone, precipitating airway obstruction during recovery from anesthesia. The combination of anesthetic effects and immature respiratory control place the micropremie at risk for central and obstructive apnea for a prolonged period of time during recovery from anesthesia.

Not surprisingly, apnea occurs commonly after anesthesia and surgery in preterm infants.^14,15 Like apnea of prematurity, postoperative apnea may be central, obstructive, or mixed in origin.¹⁶ The term postoperative apnea usually means prolonged apnea (greater than 15 seconds) or brief apnea accompanied by bradycardia (heart rate 80 beats per minute or less). Postoperative apnea typically occurs as a cluster of episodes over several minutes, with minutes of normal breathing in between the clusters. Bradycardia may occur with apnea, usually beginning at the onset of apnea and not in response to hypoxia. Arterial oxygen desaturation usually follows the apnea, although many apneic episodes may not have any associated desaturation. Arterial desaturation is worse with obstructive apnea than with central apnea.¹⁶

Several different terms have been used to describe the age of the fetus, leading to some confusion in the literature. Gestational age, menstrual age, conceptional age, and postnatal age are all used with somewhat different meanings, even in this book, so we present here the definitions of these terms according to the American Academy of Pediatrics, Committee on the Fetus and Newborn from 2004.¹⁷ The gestational or menstrual age of the neonate is the interval from the first day of the mother’s last menstrual cycle until birth of the fetus. The postmenstrual age is the sum of the menstrual age and the postnatal age. The conceptional age is defined as the interval between conception and birth, although the former is generally unknown. The postconceptional age (PCA) is the sum of the conceptional age and the postnatal age. Postconceptual age actually refers to a concept, not conception, but this term has been used interchangeably with postconceptional age in the apnea literature in anesthesia. The postnatal (or chronological) age is the age of the infant since birth. Controversy over this terminology continues, as some argue that the menstrual age overestimates the “in utero” age of the fetus because 10 to 14 days may lapse between the onset of menses and conception. On the other hand, terms such as conceptional (and postconceptional) age are imprecise because the date of conception is usually unknown, and thus these terms are not recommended.¹⁷

The incidence of postoperative apnea depends on PCA, hematocrit, and the type of surgical procedure (Fig. 35-2; see also Fig. 4-7 and E-Fig. 4-5).^14–1618 The most significant risk factor is the PCA; the lesser the PCA, the greater the risk, with the incidence of postoperative apnea in the micropremie greater than 50%.^14,15 Postoperative apnea can occur in the micropremie even without a history of apnea of prematurity.¹⁴ Anemia (hematocrit less than 30%) and younger gestation increase the risk of apnea for a given PCA.^15,18

FIGURE 35-2 Time from the end of anesthesia to the last episode of postoperative apnea in prematurely born infants. (r² = 0.49.)

(Redrawn with permission from Kurth CD, Spitzer AR, Broennle AM, et al. Postoperative apnea in preterm infants. Anesthesiology 1987;66:483-8.)

Postoperative apnea usually begins within an hour of emergence from anesthesia.¹⁴ In the micropremie, it can continue to occur up to 48 hours postoperatively, despite the elimination of anesthetic agents (see Fig. 35-2). In fact, postoperative apnea can occur after surgery with desflurane- or sevoflurane-based anesthetics, or even after surgery for which a regional anesthetic was administered and no general anesthetic drugs were used.^19,20 Postoperative apnea is more common after major procedures, such as a laparotomy, compared with peripheral surgical procedures, such as inguinal hernia repair. These observations indicate that the neurohormonal response to surgery and postoperative pain may play an important role in the origins of postoperative apnea. Management of postoperative apnea includes close observation with a cardiorespiratory monitor and pulse oximeter, administration of intravenous (IV) methylxanthines (e.g., caffeine, theophylline),²¹ and prevention of anemia or hypovolemia. Nasal CPAP or tracheal intubation and mechanical ventilation may be required for several days postoperatively if these measures fail.

N7-methylation of theophylline (or aminophylline) to produce caffeine is well-developed in the neonate, whereas oxidative demethylation (CYP1A2) responsible for caffeine metabolism is deficient. Theophylline is effective for the management of postoperative apnea in the preterm neonate, in part because it is a prodrug of caffeine, which is effective in controlling apnea in this age group and can only be cleared slowly by the immature kidney. Consequently, the half-life of caffeine is ∼72 hours in the extreme premature neonate, which decreases to 4 to 6 hours by 6 months of age. Clearance increases from 0.004 L/kg/hr in the premature neonate to 0.119 L/kg/hr by 6 months.^22–26 Although therapeutic drug monitoring is not required, target concentrations 5 to 20 mg/L are considered therapeutic.²⁷ One study suggests a loading IV or oral dose of 10 mg/kg followed by 2.5 mg/kg by mouth, once daily.²⁶

Cardiovascular System

The micropremie remains at greater risk of cardiovascular collapse during anesthesia and surgery than does the full-term infant for several reasons. The fetal heart differs from the infant heart in that it has more connective tissue, less organized contractile elements, and increased dependence on extracellular calcium concentration. In addition, the less compliant fetal heart has a flatter Frank-Starling curve and is less sensitive to catecholamines because of near-maximal baseline β-adrenergic stimulation (see Chapter 16).^28,29 Consequently, cardiac output depends more on heart rate in the micropremie than it does in the term neonate. The increased resting heart rate in the micropremie also does not permit cardiac output to increase to the same extent as in an infant or child. The micropremie has a small absolute blood volume (Table 35-2). Therefore, relatively little blood loss during surgery can cause hypovolemia, hypotension, and shock. Because autoregulation is not well developed in the micropremie, the heart rate may not increase with hypovolemia, and blood flow and oxygen delivery to the brain and heart may decrease with relatively little blood loss.³⁰ Anesthesia blunts baroreceptor reflexes in the micropremie, further limiting the ability to compensate for hypovolemia.³¹ The combination of limited ventricular stroke volume reserve, an increased heart rate, small blood volume, and limited autoregulation predispose the micropremie to cardiovascular collapse during major surgery.

Failure of the ductus arteriosus to close in the micropremie further increases this risk. A patent ductus arteriosus (PDA) promotes pulmonary hypertension and congestive heart failure. Changes in systemic or pulmonary vascular resistance alter the direction of flow through the PDA or the foramen ovale.³² Increased pulmonary vascular resistance predisposes to right-to-left shunting that worsens with hypoxia, hypercarbia, acidosis, and hypothermia. Paradoxical embolism is another concern.³³ Fluid restriction and diuretic therapy, often used to treat congestive heart failure from left-to-right shunting through a PDA, further increase the risk of hypotension during surgery. In contrast to full-term neonates, the success of inhaled nitric oxide in the micropremie with hypoxic respiratory failure and pulmonary hypertension remains unclear.^34–36

Neurologic Development

Although mortality in extremely preterm infants has improved over the years, many survivors experience cognitive impairment and long-term disability.^37,38 Regions of the central nervous system develop at different times during gestation; consequently, the impact of premature birth on the central nervous system (CNS) depends on gestational age at birth and the severity of cardiovascular, respiratory, and other postnatal stressors. The area of the brain most susceptible to injury in the micropremie is the periventricular white matter.³⁸ The white matter consists of preoligodendrocytes, astrocytes, and neuronal axons. Late in the second trimester (24 to 27 weeks gestation), preoligodendrocytes and astrocytes multiply tremendously and most cortical and subcortical structures begin to develop.³⁸ During this period, the periventricular white matter is particularly susceptible to neurologic injury. The periventricular white matter is perfused by arteries penetrating from the cortical surface and by lenticulostriate arteries from the circle of Willis. As a result, the periventricular white matter is a “watershed region” and susceptible to poor perfusion and hypoxic-ischemic injury during conditions of hypotension, reduced cardiac output, hypoxemia, and hypocarbia.

Neural pathways allowing for perception of pain develop during the first, second, and third trimesters (see Chapter 43).³⁹ During the first trimester, peripheral sensory receptors and spinal reflex arcs develop that lead to the presence of a “withdrawal reflex” to non-noxious stimuli. Neurons that transmit nociception appear in the dorsal root ganglia at 19 weeks gestation, and afferent neurons from the thalamus reach the cortical subplate and cortical plate between 20 and 24 weeks gestation. However, it is not until early in the third trimester (29 weeks) that pathways between the thalamus and somatosensory cortex are functional. Significant controversy exists regarding the exact gestational age at which perception and memory of pain occur. Nevertheless, our approach in the micropremie is to administer anesthesia during surgery and provide pain management postoperatively.

Long-Term Neurologic Complications of Prematurity

Long-term neurologic and developmental disabilities remain common in the micropremie and include cerebral palsy, cognitive deficits, behavioral abnormalities, as well as hearing and visual impairment. In one cohort of ELBW infants, only 25% were classified as normally developed at 5 years of age, whereas 20% exhibited major disabilities.³⁷ Brain magnetic resonance imaging (MRI) identifies a spectrum of abnormalities. The most common abnormality is diffuse high signal intensity on T2-weighted imaging in the periventricular cerebral white matter. Diffusion-weighted imaging shows increased apparent diffusion coefficient values, indicative of increased water content and delayed white matter maturation, suggesting ischemia-reperfusion injury in periventricular white matter, which has activated microglia and damaged preoligodendrocytes.^38,40 Damage to preoligodendrocytes impairs myelination of cerebral white matter axons and accounts for many of the fine motor, speech, and cognitive deficits. On MRI, tissue volumes in the basal ganglia, corpus callosum, amygdala, and hippocampus are reduced and correlate with smaller full-scale, verbal, and performance IQ scores.⁴¹ Collectively, these MRI findings indicate that different regions of the brain vary in their susceptibility to injury during development and that such injuries lead to specific long-term disturbances in neurocognitive function.

Intraventricular Hemorrhage

Intraventricular hemorrhage (IVH) occurs in as many as one-third of micropremie infants. The severity of IVH, as defined by head ultrasound, is graded as follows:

Although micropremie infants with grade 3 or 4 IVH are more likely to exhibit severe long-term neurocognitive sequelae, even micropremie infants with grade 1 and 2 IVH display poorer neurodevelopmental outcomes compared with those without IVH.⁴² Early onset of IVH appears during the first day of life. Risk factors include fetal distress, vaginal delivery, reduced Apgar scores, metabolic acidosis, severe hypercapnia, and the need for mechanical ventilation.^43,44 Late onset of IVH appears days to weeks after birth. Risk factors include respiratory distress syndrome, seizures, pneumothoraces, hypoxemia, acidosis, severe hypocarbia, and the use of vasopressor infusions.⁴³ Rapid fluctuations in cerebral blood flow, cerebral blood volume, and cerebral venous pressure appear to play a role in the development of IVH.⁴⁵ Factors that may decrease the incidence and severity of IVH include administration of antenatal glucocorticoids, or indomethacin.

Retinopathy of Prematurity

Retinopathy of prematurity (ROP) occurs in approximately 50% of ELBW infants, with the incidence being inversely proportional to birth weight and gestational age (see Chapter 32).⁴⁶ Although the pathogenesis of ROP is not completely understood, extremes in arterial oxygenation (hypoxia or hyperoxia)⁴⁷ and exposure to bright light appear to play a role.⁴⁸ One theory holds that the combination of hyperoxic vasoconstriction of retinal vessels (also known as vaso-obliteration), induction of vascular endothelial growth factor, and free oxygen radicals damage the spindle cells in the retina.⁴⁹ A Cochrane review concluded that liberal oxygen delivery to a preterm infant is more harmful to the retina than restrictive oxygen delivery, although the data reviewed failed to specify the optimum blood oxygen concentrations that should be delivered.⁵⁰ Evidence points to additional factors in the pathogenesis of ROP, including genetic polymorphisms⁵¹ and antenatal and neonatal exposure to inflammation.⁵²

ROP appears to be multifactorial in origin and oxygen tension is just one of many contributory factors. During anesthesia, our goal is to deliver the minimum inspired oxygen concentration that provides oxygen saturations between 90% and 94% and to avoid significant fluctuations in oxygen saturations. It should be noted however that ROP has occurred in children with cyanotic congenital heart disease⁵³ and that no anesthesia-associated cases have been reported over the past 25 years. Nevertheless it is reasonable to aim for saturation values in the ranges described here.

Temperature Regulation

The micropremie is susceptible to hypothermia. Heat loss in children occurs by four possible routes: radiation (39%), convection (34%), evaporation (24%) and conduction (3%). In the micropremie, evaporative heat loss and insensible fluid loss are increased because the epidermis has less keratin.⁵⁴ Conductive and convective heat losses are also increased because the micropremie has little subcutaneous fat for insulation and a large surface area to mass ratio. Thermal regulation is not well developed in the micropremie. Nonshivering thermogenesis, which depends on brown fat stores, is decreased and regulation of skin blood flow is less efficient.⁵⁵ During anesthesia, measures should be undertaken to minimize radiation and convective heat loss by warming the operating room (OR) to 78° F to 80° F (25.5° C to 26.6° C) before the neonate arrives, and minimize convective heat loss during transport (i.e., use a thermoneutral incubator). Using a warming pad on the operating table reduces conductive heat loss; use of overhead heat lamps reduces radiant heat loss; and keeping the skin dry reduces evaporative heat loss. The most effective means for warming is a forced-air warmer. Temperature should be carefully monitored as overheating the infant may readily occur.

Renal and Metabolic Function

In the micropremie, kidney function is decreased as a result of fewer nephrons and smaller glomerular size.⁵⁶ Glomeruli continue to form postnatally until approximately 40 days.⁵⁷ During this period, reduced cardiac output, hypotension, and nephrotoxic drugs may inhibit glomerular growth and development. Creatinine concentrations depend on production that is reduced in micropremies with limited muscle mass, and on excretion that is reduced because of immature renal function. Baseline plasma creatinine concentrations increase with increasing prematurity and remain increased until 3 weeks of age.⁵⁸ In addition, the normal increase in creatine clearance in term infants occurs more slowly in the micropremie. Creatinine concentrations in the first few days after birth are increased and reflect maternal transplacental transfer.⁵⁹ It is for this reason that antibiotic dosing must be adjusted to take renal immaturity into consideration, so as not to administer excessive doses that might result in ototoxicity.⁶⁰

Very preterm infants easily become hyponatremic because of reduced proximal tubular reabsorption of sodium and water, and reduced receptors for hormones that influence tubular sodium transport. As many as one-third of ELBW neonates develop hyponatremia.⁶¹ Frequent assessment of sodium and free water requirements is important during critical illness. Increased plasma potassium concentrations occur in preterm infants during the first few days after birth. The increase results from a shift in potassium from the intracellular to extracellular space.⁶² These increases are greater as gestational age and birth weight decrease.⁶³ Reduced cardiac output and urine output may further increase serum potassium concentrations and predispose to cardiac arrhythmias.⁶⁴

Glucose Regulation

The micropremie is at risk for both hypoglycemia and hyperglycemia. Decreased glycogen and body fat predispose to fasting hypoglycemia, whereas decreased insulin production with infusion of dextrose predisposes to hyperglycemia.^65,66 Glucose production is poorly regulated within a large range of glucose and insulin concentrations. The micropremie is also relatively insulin resistant and requires a greater infusion rate of insulin to achieve normoglycemia.⁶⁷ The use of total parenteral nutrition and glucocorticoids places the micropremies at increased risk for hyperglycemia.

Glucose and the Brain

Multiple animal models and clinical studies implicate hyperglycemia as detrimental to the adult brain during global and focal ischemia.⁶⁸ In contrast, hyperglycemia in neonates appears to protect the brain from ischemic damage.^69–71 Studies in both neonatal rat and pig hypoxia-ischemia models observed less brain damage with greater glucose concentrations. Many mechanisms exist for this strikingly different outcome between neonates and adults.⁷² Relatively mild hypoglycemia is known to cause brain damage in preterm infants.⁷³ Micropremies with critical illness are especially prone to hypoglycemia because they contain limited stores of glucose and consume glucose anaerobically. Thus the administration of dextrose-containing fluids (carefully controlled with an infusion pump so as to minimize wide fluctuations in glucose values) and close monitoring of blood glucose concentrations is vital during anesthesia. Mild or moderate hyperglycemia during surgery is best managed by reducing the rate of infusion of dextrose-containing solutions and not administering insulin, with its attendant risk of hypoglycemia.

Hepatic and Hematologic Function

Immature hepatic function leads to a reduction in many hepatic proteins important for drug metabolism. In addition, reduced albumin synthesis decreases albumin concentrations compared with term neonates (see Fig. 6-6), thus enhancing the “free” (unbound) concentration of anesthetic drugs that are highly bound to albumin (see Chapter 6). The micropremie is at particular risk for spontaneous liver hemorrhage.^74,75 This occurs most commonly during laparotomy for necrotizing enterocolitis (NEC), is associated with large IV fluid resuscitation, and is difficult to control surgically. Recombinant factor VIIa has been used to stop liver hemorrhage when administration of other blood products has been unsuccessful.⁷⁶

The ideal hematocrit level for the micropremie remains controversial. In the micropremie with reduced oxygen saturations and cardiac output, tissue oxygen delivery will be maximized by maintaining the hematocrit between 44% and 48%. In a randomized study of liberal versus restrictive transfusion in neonates between 500 and 1300 grams, intraparenchymal brain hemorrhage, periventricular leukomalacia, and episodes of apnea occurred more frequently in the restrictive transfusion group.⁷⁷ The risks of blood transfusion in the micropremie must be balanced against the benefits of improved oxygen delivery and fewer medical complications.

Thrombocytopenia (platelet count less than 150,000/mm³) occurs in as many as 70% of micropremies.⁷⁸ Although the etiology of thrombocytopenia is often unknown, pathophysiologic processes such as sepsis, disseminated intravascular coagulation, and NEC are common causes. Preoperative evaluation should include a recent platelet count and the availability of platelets for major procedures.

Anesthetics and the Immature Brain

Research in immature animals indicates that anesthetics are both neuroprotective and neurotoxic. Inhalational anesthetics protect against hypoxic-ischemic injury in neonatal pigs and rats.^79–81 The anesthetic must be administered before and during the ischemic event at a concentration of 1 MAC (minimal alveolar concentration) to be effective. Thus for surgery in which there is a risk of brain ischemia, use of an inhalational anesthetic may afford some advantage over IV agents. Cardiac surgery, ventricular shunt insertion, and vein of Galen embolization represent examples of procedures that are performed in preterm infants and that carry a risk of brain ischemia. The MAC for sevoflurane has not been established in preterm infants and many sick preterm infants cannot tolerate even relatively modest concentrations of potent anesthetic agents.

Of particular concern are the reports in immature rats and other animals, including primates, that prolonged exposure to commonly used anesthetics, such as isoflurane, ketamine, and midazolam, induces apoptosis in many regions of the brain (see Chapter 23).^82,83 In rodents, exposure for at least 2 hours at 1 MAC of isoflurane produces apoptosis. A combination of isoflurane, midazolam, and nitrous oxide produces more neuronal degeneration than isoflurane or midazolam alone; nitrous oxide alone is not neurotoxic. When affected rats matured to adulthood, neurocognitive impairment was detected.⁸² The neurotoxicity is brain-region specific and very dependent on the developmental age of the rodent. Rats are most sensitive to the neurotoxic effects of anesthetics on postnatal day 7, more so than on postnatal day 4 or beyond postnatal day 10.⁸⁴ The most susceptible age in rats, 7 days, corresponds to human brain development around mid-gestation. This suggests that if this phenomenon applies to humans, the preterm infant could potentially be more susceptible to anesthetic neurotoxicity than is the full-term infant.

The mechanism for the neurotoxicity appears to be attributable to the neurotransmitters glutamate and γ-aminobutyric acid, which act as trophic factors in the developing brain.⁸⁵ In the immature brain, these trophic factors promote synaptic growth and plasticity and are obligate for neuronal survival. The inhaled anesthetics, ketamine, and midazolam exert their anesthetic effects by altering synaptic transmission through blockade of glutamate and γ-aminobutyric acid receptors. In the immature brain, this blockade also precipitates neuronal cell death by apoptosis.⁸⁶ In contrast, several anesthetics and medications may protect against apoptosis (see Chapter 23). A confounding factor is that neurodegeneration and apoptosis is a normal developmental phenomenon in the maturing fetal brain. Furthermore, anesthesia-induced neuronal cell death in neonatal animals may not directly translate into long-term neurologic abnormalities. Indeed, evidence suggests that sevoflurane-induced cognitive impairment, in the form of short-term memory deficiency in neonatal rodents, is offset by delayed exercise.⁸⁷ Moreover, immature animals that undergo painful procedures without anesthesia experience neuronal degeneration.^88,89 Preterm infants who receive anesthesia and sedation for painful procedures experience less morbidity and mortality than those who do not.⁹⁰ Curiously, the combination of surgery and anesthesia in neonatal rats produces more apoptosis than either intervention alone suggesting that in this model, anesthetics are neither neuroprotective themselves nor do they offset the apoptotic effects of surgery.⁹¹ In summary, the neurodegeneration precipitated by inhaled anesthetics, ketamine, and benzodiazepines depends on developmental age, brain region, and duration of exposure. Based on the animal models, the micropremie exposed to several hours of large concentrations of inhaled agents with nitrous oxide and midazolam is potentially at risk, as is the micropremie exposed to surgery with insufficient anesthesia. Thus our approach at the present time for emergency surgery is to use small concentrations of inhaled agent with opioids and regional anesthesia whenever possible.

Of even greater concern may be the sedatives that are administered for prolonged periods of time in the intensive care unit (ICU), although one study found “no evidence of an association between dose and duration of sedation and/or analgesia drugs given during the preoperative, intraoperative, and postoperative period and major adverse developmental outcomes” in children undergoing repair of congenital heart disease in the first 6 weeks of life.⁹²

Inhalation Anesthetic Agents

The MAC defines the anesthetic depth for inhaled agents at which 50% of patients respond to a painful stimulus with movement; this measure allows comparison of the effects of inhaled anesthetics at equipotent doses. The MAC of isoflurane in the micropremie (less than 32 weeks PCA) is approximately 20% less than that in full-term neonates (E-Fig. 35-1; see also Fig. 6-16), and that at equipotent doses of isoflurane (1 MAC), systolic arterial pressure decreased similarly in all age-groups, 20% to 30%.⁹³ Sevoflurane affords a rapid induction and emergence from general anesthesia. Desflurane is contraindicated for induction of anesthesia but is widely used for maintenance of anesthesia administered through an ETT. However, desflurane causes more airway irritability than isoflurane or sevoflurane, and as a result, it is not recommended for infants with severe BPD. Desflurane, sevoflurane, and isoflurane decrease arterial blood pressure in a dose-dependent manner, possibly through decreasing the systemic vascular resistance or by myocardial depression. One possible mechanism to explain the myocardial depression is that the baseline ionized calcium concentrations in preterm infants, especially critically ill neonates are reduced.^94,95 Because inhalational anesthetics block the calcium channels,⁹⁶ and because the neonatal heart depends on the plasma ionized calcium for contractility to a greater extent than do the hearts of older children,⁹⁷ preterm infants may be more susceptible to the cardio-depressant effects of inhalational anesthetics.

E-FIGURE 35-1 The minimal alveolar concentration (MAC) for isoflurane in preterm infants, full-term infants, children, and adults. (Note data are plotted in postconceptional years.)

(Reprinted with permission from LeDez KM, Lerman J. The minimum alveolar concentration [MAC] of isoflurane in preterm neonates. Anesthesiology 1987;67:301-7.)

Nitrous oxide is not routinely used in the micropremie for several reasons. First, nitrous oxide must be delivered in inspired concentrations ranging from 50% to 75% to reduce the MAC of other agents; therefore, its role in micropremies, a group often requiring supplemental oxygen, is limited. Second, because of its blood gas solubility, nitrous oxide rapidly enters air-filled cavities; therefore it is not recommended for use in infants with bowel obstruction, NEC, pulmonary interstitial emphysema, or pneumothoraces, which are common disorders in micropremies.⁹⁸ Third, in neonatal and young rats, nitrous oxide demonstrates no antinociceptive effects, which contrasts its antinociceptive effect in adolescent and adult rats.⁹⁹ This observation requires validation in humans.

Intravenous Agents

IV agents include opioids, benzodiazepines, barbiturates, propofol, ketamine, and dexmedetomidine. Fentanyl possesses analgesic and sedative properties, however, it does not reliably produce unconsciousness or amnesia and, by itself, is not considered an anesthetic in children or adults. Nevertheless, the use of fentanyl as an anesthetic has been justified in preterm infants because they were deemed to be inherently amnestic by virtue of their age, even though the age at which “consciousness” and memory occurs is unknown. Preterm infants (less than 1500 g) who receive IV fentanyl (30 to 50 µg/kg) and pancuronium for ligation of a PDA exhibit remarkable hemodynamic stability, with only a 5% decrease in blood pressure.¹⁰⁰ Another study examined the dose response of 25 neonates undergoing a variety of thoracic and abdominal procedures. A dose of 10 to 12.5 µg/kg administered together with a muscle relaxant produced hemodynamic stability for 75 minutes.¹⁰¹ Hypertension and tachycardia did not occur with skin incision, suggesting that analgesic concentrations necessary for surgery are achieved with this dose of fentanyl.

The pharmacokinetics of fentanyl (30 µg/kg) in preterm infants yielded plasma concentrations that remained constant for up to 120 minutes, indicating a reduced clearance.¹⁰² The elimination half-life of fentanyl ranged from 6 to 32 hours in preterm infants, significantly greater than the 2- to 3-hour half-life observed in children and adults.¹⁰² The clearance of fentanyl is 7 mL/min/kg at 25 weeks, 10 mL/min/kg at 30 weeks, and 12 mL/min/kg at 35 weeks postmenstrual age.¹⁰³ These studies demonstrated that the half-life and volume of distribution of fentanyl are increased, whereas the clearance is reduced in preterm infants compared with adults.¹⁰⁴ These changes may be explained by immature CPY450 3A4, reduced proteins, immature kidneys, and a patent ductus venosus. In a subset of infants with increased intra-abdominal pressure (after repair of a gastroschisis or omphalocele), the elimination half-life of fentanyl is 1.5- to 3-fold greater than that in other infants of the same age.¹⁰⁴ This likely results because increased intraabdominal pressure decreases hepatic blood flow, the rate limiting step in the metabolism of drugs with a large hepatic extraction ratio, such as fentanyl.¹⁰⁴ The increased volume of distribution decreases the initial plasma concentration of fentanyl compared with that in adults. These pharmacokinetic differences, combined with an increased propensity to apnea, serve to prolong analgesia, as well as prolong respiratory depression, increase the risk of postoperative apnea, and slow recovery of consciousness. In the micropremie, mechanical ventilation may be required for several days after large doses of fentanyl.

Similarly, the elimination half-life of morphine is markedly prolonged in preterm infants compared with that in children and adults.^105–108 The elimination half-life of morphine ranges from 6 to 16 hours in the micropremie, compared with 2 to 4 hours in the adult. We prefer fentanyl instead of large-dose morphine (2 to 3 mg/kg) for anesthesia because it has fewer hemodynamic adverse effects.

Remifentanil, a relatively new synthetic, short-acting opioid, is rapidly inactivated by plasma and tissue esterases and, because of its short half-life, is administered by continuous infusion. The half-life of remifentanil in adults is 3 to 4 minutes, independent of the duration of infusion, and similar to that in infants or children.¹⁰⁹ A multicenter study that compared halothane and remifentanil for maintenance of anesthesia in infants undergoing pyloromyotomy showed similar intraoperative hemodynamic stability intraoperatively with the two techniques, but significantly fewer “new onset apneas” with remifentanil compared with halothane.^110,111 Interestingly, the most rapid clearance of remifentanil was in infants and children younger than 2 years of age, thus allowing an intense opioid effect intraoperatively that rapidly dissipates on terminating the infusion.¹¹² Remifentanil has been used to provide anesthesia in infants weighing 400 to 580 grams with apparent good hemodynamic stability.^113,114 A study examining cord blood from preterm infants found high nonspecific esterase activity, comparable to that of term infants, thus suggesting that preterm infants should be able to rapidly metabolize remifentanil.¹¹⁵

Ketamine, a phencyclidine derivative, affords several advantages compared with inhaled and other IV agents. It provides analgesia, amnesia, and unconsciousness yet minimally depresses cardiovascular function (Fig. 35-3).¹¹⁶ However, ketamine anesthesia depresses ventilation and airway reflexes, which predisposes to airway obstruction, apnea, and gastric aspiration. Thus we recommend the use of an ETT when ketamine is used for surgical procedures in the micropremie. In the setting of brief painful procedures, IV ketamine can be used as an anesthetic without an ETT.¹¹⁷

FIGURE 35-3 A, Changes in systolic arterial pressure (SAP) in preterm infants after anesthesia with either isoflurane, halothane, fentanyl, or ketamine, and after surgical incision. B, Changes in mean arterial pressure (MAP) in premature infants after anesthesia with either isoflurane, halothane, fentanyl, or ketamine, and after surgical incision.

(Reprinted with permission from Friesen RH, Henry DB. Cardiovascular changes in preterm neonates receiving isoflurane, halothane, fentanyl, and ketamine. Anesthesiology 1986;64:238-42.)

Other IV agents include thiopental, propofol, and benzodiazepines. These agents induce loss of consciousness but possess less analgesia than ketamine. Thiopental is a short-acting barbiturate primarily used for the induction of anesthesia. The micropremie requires less thiopental for induction than does the infant (2 to 3 mg/kg vs. 5 to 6 mg/kg, respectively), a relationship similar to the MAC of isoflurane.¹¹⁸ In the micropremie, we only use thiopental for neurosurgical procedures involving increased intracranial pressure. However, thiopental is no longer available in the USA. Propofol is primarily used to induce anesthesia and has largely replaced thiopental for this purpose. A word of caution is needed regarding the use of propofol for induction of anesthesia in neonates. Several reports highlight episodes in otherwise stable infants of protracted hypotension and low cardiac output state that were associated with hypoxia after propofol boluses (1 to 3 mg/kg IV). The mechanism underlying these responses remains unclear although acute pulmonary hypertension with reversion to persistent fetal circulation remains a strong possibility.^119,120 In our experience, the micropremie can be anesthetized with a propofol infusion (50 to 200 µg/kg/min) supplemented with fentanyl as needed for analgesia. The selection of infusion pumps that allow for delivery of small volumes accurately is vital. The infusion rate of propofol in these small infants must be carefully and meticulously checked, as a 10-fold overdose of propofol has been reported, with a successful recovery.¹²¹ Recovery from propofol anesthesia is delayed in micropremies compared with term infants, because micropremies have both less fat and muscle tissue to redistribute the drug, and reduced clearance. In pediatric ICUs, propofol infusions have been implicated in unexpected deaths (propofol infusion syndrome).¹²² Until the safety of long-term administration of propofol has been examined in preterm infants, other alternatives for prolonged sedation should be considered.

Benzodiazepines, such as midazolam and diazepam, have been used in the neonatal intensive care unit (NICU) for sedation. As with thiopental and propofol, these drugs do not provide analgesia and are not recommended as the sole anesthetic for surgery. However, the combination of a benzodiazepine and opioid provides complete anesthesia for surgery. Midazolam clearance is markedly decreased in the micropremie when compared with the term neonate or infant, and will be further prolonged in the setting of decreased liver function.¹²³ Midazolam can cause systemic hypotension, depress ventilation, and impair airway reflexes in preterm infants. The hypotension caused by midazolam is greater in the presence of fentanyl; thus both drugs must be titrated in small doses when administered concomitantly.¹²⁴ One study noted an 8% to 23% decrease in arterial pressure after a bolus of 0.1 mg/kg of midazolam in preterm infants.¹²⁵

Exploratory Laparotomy for Necrotizing Enterocolitis

Necrotizing enterocolitis, a life-threatening condition mainly afflicting preterm neonates, occurs in about 5% of ELBW infants (see Chapter 36).¹²⁸ Although NEC may be treated medically, the micropremie with NEC is more likely to require surgery; mortality ranges from 10% to 50%.^129–131 The pathogenesis of NEC is incompletely understood; intestinal mucosal ischemia is thought to play a key role. Other key contributing factors include inflammation of bowel mucosa, alterations in normal intestinal flora by antibiotic therapy, gastric alkalinity, low systemic cardiac output, and red blood cell transfusion in the preceding 48 hours.^132–135 Early signs of NEC include feeding intolerance, increased work of breathing, lethargy, and temperature instability; later signs include hypotension, abdominal distention, apnea, thrombocytopenia, coagulopathy, and multisystem organ failure. Classic radiographic findings include gas in the intestinal wall (pneumatosis intestinalis) and biliary tract, and free air within the abdomen. Indications for surgical exploration include the presence of perforation or continued clinical deterioration despite medical management (Fig. 35-4).

FIGURE 35-4 Early necrotizing enterocolitis with bowel perforation. Note that the perforation was diagnosed early and that there is soiling of the peritoneum but there does not appear to be any dead bowel. This type of perforation is generally associated with a positive outcome.

Surgical management of the micropremie with NEC involves either initial primary peritoneal drainage or a laparotomy with resection of necrotic bowel. Primary peritoneal drainage requires a small surgical incision and fewer anesthetic requirements and can be performed at the bedside. Some infants who undergo initial peritoneal drainage will subsequently require laparotomy if their condition worsens. Peritoneal drainage has generally been favored in the smaller and more unstable preterm infants for logistical reasons; thus the ability to compare the two management strategies has been difficult because of multiple confounding variables.^128,136 A prospective randomized, multicenter trial comparing laparotomy and peritoneal drainage in infants weighing less than 1500 g with perforated NEC found no difference in survival, development of short-gut syndrome, or length of hospital stay between the two approaches.¹³⁷

NEC is a surgical emergency, and preoperative preparation in a timely fashion is vital. NEC predisposes to hypovolemia, cardiovascular and respiratory failure, capillary leak syndrome, disseminated intravascular coagulation, and hypoglycemia. It is important to prepare for all these sequelae. We have albumin, fresh frozen plasma, platelets, and packed red blood cells available; calcium gluconate, dopamine, and epinephrine are prepared for treatment of hypotension and reduced cardiac output. Because pulmonary function may deteriorate, we ensure the availability of sophisticated ventilators for increased mechanical ventilatory support. Newer anesthesia machines provide modes of ventilation similar to the ventilators used in the ICU. Coagulopathy and thrombocytopenia increase bleeding and necessitate blood product administration during surgery. Before transporting the micropremie to the OR, a transfer note is written to include the current medications, size and location of IV catheters, and dextrose infusion rate. We also obtain a current plasma glucose concentration and assess the ETT for size, equality of breath sounds, how well it is secured, and when it was last changed.

Vascular access remains critical for this surgery. At least two venous cannulas and an arterial catheter should be considered for optimal care. Usually the venous cannulas are 24-gauge catheters. In-line filters designed to prevent air within tubing, easily become blocked when one attempts to transfuse blood through them. Packed cells, even if they are reconstituted with plasma, do not flow rapidly through 24-gauge catheters; therefore it is important to avoid falling behind in fluid administration in these infants, so that there is no need to force a catch-up to stave off hypotension. A central venous catheter facilitates the delivery of infusions, such as dopamine or epinephrine. In our institution, many preterm infants have peripherally inserted central catheters (PICCs), allowing for delivery of vasopressors. These catheters are not well-suited for rapid delivery of anesthetic drugs or blood, or for measurement of central venous pressure. An arterial catheter provides continuous blood pressure monitoring and the ability to sample blood gases, electrolytes, complete blood cell count (CBC), platelet count, and prothrombin time/partial thromboplastin times (PT/PTT). In recent years, automated blood pressure cuffs have improved enormously in their ability to measure blood pressure non-invasively in young infants, although cases still occur in which they fail to measure arterial pressure during this surgery. A Foley catheter allows for the assessment of urine flow. A nasogastric tube should be present. Time spent by the anesthesia and surgical team to place these lines must be weighed against the need for urgent laparotomy.

General endotracheal anesthesia with neuromuscular blockade remains the anesthesia technique for NEC surgery. NEC increases the risk for aspiration. Tracheal intubation may be achieved by “awake intubation” or by rapid-sequence anesthetic induction. We prefer an awake intubation in the micropremie because effective cricoid pressure is difficult to apply, a very small force may distort the airway in the infant rendering intubation difficult,¹³⁸ and arterial desaturation occurs rapidly during apnea. After the administration of IV atropine, awake intubation is performed using an oxyscope, a laryngoscope with an oxygen port to allow for oxygen delivery during direct laryngoscopy (see E-Fig. 12-12, A and B). Awake intubation should be completed efficiently and rapidly in less than 15 seconds. To accomplish this, all equipment and monitors should be prepared, including an ETT with a “hockey stick” bend using a stylet. After preoxygenating the infant and administering 10 to 20 µg/kg IV atropine, the infant’s arms are brought up by an assistant and held against the head (hands pointing to the anesthesiologist) with the elbows adjacent to the ears. This secures the child’s arms and shoulders, preventing the infant from laterally rotating the neck, raising the shoulders, or moving the arms during laryngoscopy. Once the infant is positioned, a #1 straight blade is inserted into the mouth at the right commissure and the tip is advanced toward the glottic opening in one smooth motion. When the infant gags, the ETT, which is being held in the other hand, is poised to pass through the cords. Once the position of the tube is confirmed with the presence of carbon dioxide on the capnogram, the assistant immediately administers a predetermined dose of IV propofol to rapidly induce anesthesia. Inhalational anesthesia and muscle relaxant, as indicated, should be administered while the ETT is taped in place. Chest auscultation in the axillae will confirm a properly placed ETT.

The anesthetic regimens that we prefer for maintenance include (1) IV fentanyl (5 to 10 µg/kg), an inhaled anesthetic (such as isoflurane, sevoflurane, or desflurane), and neuromuscular blocking drug (balanced technique), or (2) IV fentanyl (20 to 50 µg/kg), IV midazolam (0.1 mg/kg), and a neuromuscular blocking drug (high-dose opioid technique). We select the high-dose opioid technique for hemodynamically unstable infants. Pancuronium is an ideal neuromuscular blocking drug when combined with large doses of fentanyl because of its anticholinergic properties, but should be avoided in infants with renal dysfunction. If hypotension persists despite a trial of 10 to 20 mL/kg of IV fluid, we begin an infusion of epinephrine (0.02 to 0.1 µg/kg/min) or dopamine (5 to 20 µg/kg/min). Hypocalcemia during administration of citrated blood products may contribute to hypotension, and requires replacement with either calcium chloride or calcium gluconate (see Chapter 10). When severe shock persists despite IV fluid resuscitation, calcium, and inotropic support, rescue treatment with “stress dose” glucocorticoids may be beneficial. Treatment with hydrocortisone and dexamethasone has been effective in improving arterial pressure in low–birth-weight infants with refractory hypotension.^139,140

Intraoperative fluid management in micropremies should begin with continuing the solution that arrives with the infant from the NICU; usually this is a calcium- and/or glucose-containing solution. Alternately, some infants may arrive with a hyperosmolar glucose or dextrose (10%) parenteral nutrition solution. In both cases, these solutions should not be discontinued, but rather continued at the same rate (by infusion pump) or slightly less throughout the surgery to avoid reactive hypoglycemia from increased circulating insulin concentrations. There is no evidence regarding the optimal infusion rates for these solutions during anesthesia. If no solution is being infused, a balanced salt solution (e.g., lactated Ringer’s solution) could be initiated at 4 mL/kg/hr, supplemented with the same solution for third space loss (at least 10 mL/kg/hr), and replacement of blood loss. If no glucose solution is being administered, then a balanced salt solution containing glucose may be administered through a pump. Serum glucose concentrations should be monitored regularly to avoid hypoglycemia. Third space losses include evaporation and vascular leak and are replaced with an isotonic salt solution. Blood losses are replaced with packed red blood cells and fresh frozen plasma to maintain the hemoglobin greater than 10 g/dL and the PT/PTT within normal range. Platelets are administered to keep the platelet count greater than 100,000/mm³. Continuous measurement of arterial pressure and serial measurement of urine output, blood gases, CBC, platelet counts, and PT/PTT aid in the fluid replacement process. Arterial blood gas analysis helps guide ventilation and inspired oxygen concentration. Warming of the operating suite to 80° F (26.7° C), a forced air warmer underneath the infant, and warmed fluids help maintain normothermia during surgery. In preterm infants with NEC, postoperative mechanical ventilation remains the rule. Postoperative analgesia can be provided with a continuous infusion of fentanyl (1 to 3 µg/kg/hr) or intermittent doses of morphine (0.1 mg/kg every 4 to 6 hours). The time it takes for the micropremie to emerge from anesthesia depends on the anesthetic technique (balanced vs. high-dose opioid) and the need for postoperative analgesics. After a high-dose opioid technique, the micropremie may take 12 to 24 hours to emerge, as compared with several hours for a balanced technique. Remifentanil may provide an alternative, which allows an intense opioid effect that rapidly dissipates on discontinuation of the infusion^{113,114,141–143}; an alternate form of analgesia is required for postoperative pain management.

Ligation of Patent Ductus Arteriosus

Failure of the ductus arteriosus to close after birth is common in the micropremie.¹⁴⁴ A PDA may incur significant left-to-right shunting of blood, causing excess pulmonary blood flow, congestive heart failure, and respiratory failure. In fact, the diameter of the PDA may be greater than the aorta. In the micropremie with respiratory distress syndrome or persistent pulmonary hypertension, right-to-left shunting across the PDA may occur, producing cyanosis. Significant controversy exists regarding whether a PDA should be aggressively treated, the timing of therapy, and merits of medical versus surgical therapy.¹⁴⁵ Medical therapy involves the administration of a cyclooxygenase inhibitor, such as indomethacin or ibuprofen. Indomethacin therapy is less likely to close the PDA in micropremies compared with preterm infants, and more likely to produce complications, including thrombocytopenia, renal failure, hyponatremia, and intestinal perforation.¹⁴⁶ Ibuprofen is equally effective for PDA closure in the micropremie, with a reduced frequency of renal failure.¹⁴⁷ When surgery is performed by experienced teams, the incidence of major intraoperative complications is small.¹⁴⁸ However, as many as one-third of preterm infants develop severe cardiovascular instability following PDA ligation, as evidenced by systemic hypotension, pulmonary hypertension, and myocardial dysfunction.¹⁴⁹

During preoperative preparation, arterial pressure, heart rate, arterial blood gases, ventilator settings, and inspired oxygen concentration should be noted. The PDA is ligated through a left thoracotomy and requires retraction of the left lung, which decreases lung compliance, ventilation, and oxygenation. Preoperative difficulty with ventilation and oxygenation forecasts trouble in the OR. Because the aorta and pulmonary artery lie in proximity to the PDA, severe bleeding may occur abruptly and unexpectedly during the procedure. Packed red blood cells should be immediately available for transfusion. Intraoperative monitoring includes a blood pressure cuff (right arm reflects cerebral perfusion of preductal blood), continuous end-tidal carbon dioxide monitoring, and a pulse oximeter placed on digits on the right arm and a lower extremity. This will help the surgeon to confirm that the vessel about to be ligated (clipped) is in fact the ductus and not the aorta. It should be noted that with a left-sided arch, the pulse oximeter may need to be applied to the left hand instead of the right to ensure monitoring of a preductal blood vessel. Invasive monitoring of arterial pressure and blood gases is helpful in the micropremie with significant heart failure and/or lung disease, although we do not consider it a requirement for surgery. An IV catheter through which blood can be rapidly administered should be available.

The anesthetic technique of choice for PDA ligation remains fentanyl (20 to 50 µg/kg) and pancuronium (0.2 mg/kg).¹⁰⁰ Although this technique usually does not cause hypotension or bradycardia, reduction in arterial pressure after anesthetic induction may occur because of loss of sympathetic tone, especially in the setting of hypovolemia from diuretic therapy. Thus we commonly administer 5% albumin or balanced salt solution (10 mL/kg) before induction. During mechanical ventilation, mild hypoventilation and a reduced inspired oxygen concentration help to reduce pulmonary over-circulation from the PDA. However, during surgical retraction of the lung it is usually necessary to increase the ventilator inspiratory pressure setting as well as the inspired oxygen concentration. Surgical complications of PDA ligation include inadvertent ligation or laceration of the aorta or pulmonary artery. The lower extremity oxygen saturation provides a monitor of perfusion to the legs, and loss of this signal immediately after PDA ligation may indicate aortic ligation. With ligation of the pulmonary artery, oxygen saturation in both extremities and end-tidal CO₂ decrease. With successful PDA ligation, arterial diastolic and mean pressure increase, and the PDA murmur disappears.

The time until emergence from anesthesia depends on the fentanyl dose. In the micropremie with normal renal function, pancuronium will disappear in 2 to 3 hours, whereas the fentanyl (20 to 50 µg/kg) will last for at least 6 hours. Thus a trial of spontaneous ventilation and extubation is generally not planned until the following day.

Inguinal Hernia Repair

Inguinal hernias are common in preterm infants. In ELBW infants, an inguinal hernia occurs in approximately one-third of patients, whereas in full-term neonates, it occurs in 1%.¹⁵⁰ Complications related to inguinal hernias and their surgical repair include incarcerated bowel, intestinal obstruction, gonadal infarction, infection, hematoma, and recurrent hernias.¹⁵¹ Because of the risk of incarceration and bowel infarction, the hernia should be repaired as soon as the infant is medically stable.

General and/or regional anesthesia may be used for inguinal hernia repair. Induction of general anesthesia can be achieved with IV propofol or inhalation of sevoflurane. We prefer IV induction for the infant with an IV catheter in place. Indications for emergency surgery include incarceration of the bowel or gonads. When the micropremie with intestinal obstruction or incarceration requires emergency surgery, we secure the airway by awake tracheal intubation (see previous section on Exploratory Laparotomy for Necrotizing Enterocolitis). When hernia repair is performed electively, we induce anesthesia with propofol if IV access is present, or by mask with sevoflurane. A non-depolarizing muscle relaxant may be administered as needed. Cisatracurium or atracurium are preferred because their action is terminated by Hofmann elimination in plasma, whereas the action of rocuronium and vecuronium is terminated primarily by hepatic metabolism. After induction of anesthesia, the trachea is intubated and the lungs ventilated mechanically.

General anesthesia is maintained with an inhalational anesthetic agent. We avoid nitrous oxide when intestinal obstruction is present, or when the infant has complex hernias that may take several hours to repair. Preterm infants with BPD often exhibit a compensated respiratory acidosis and will have increased end-tidal CO₂ during anesthesia. In these infants the ventilation parameters are set to allow permissive hypercapnia. Mechanical ventilation with small tidal volumes (4 to 6 mL/kg), increased respiratory rates, and PEEP minimizes atelectasis and reduces the risk for lung injury. If extubation is planned after surgery, caffeine (10 mg/kg) is administered IV to reduce the risk of postoperative apnea.^21,152 IV fluids consist of lactated Ringer’s solution (4 mL/kg/hr) with dextrose (5 mg/kg/min administered by infusion pump) to maintain normoglycemia. A warm OR and forced-air warming blanket should be used to prevent hypothermia during the procedure.

Postoperative analgesia may be provided by regional analgesia or systemic analgesics; the choice depends on whether extubation of the trachea occurs immediately after surgery. Ilioinguinal and iliohypogastric nerve blocks or caudal epidural blocks may be used (see Chapter 41). We do not advocate percutaneous blockade because the local anesthetic injectate may distort the local tissues, making an already difficult surgery more difficult. Ilioinguinal and iliohypogastric nerve blocks consist of injecting 0.25% bupivacaine with epinephrine (1: 200,000) (1 mg/kg/side) around the nerve by the surgeon under direct vision. This block provides 6 to 8 hours of analgesia. Caudal epidural blockade consists of injecting 0.125% bupivacaine with epinephrine (1 : 200,000) (1 mL/kg) through the sacral hiatus, providing several hours of analgesia. Orally administered acetaminophen (10 to 15 mg/kg) can also be given, providing analgesia for several hours. We do not administer opioids to the micropremie whose trachea will be extubated immediately after surgery because of the risk of postoperative apnea.

The micropremie with BPD who requires supplemental oxygen, but who is not intubated, is an excellent candidate for regional anesthesia.^153–156 Regional anesthesia circumvents the need to intubate the trachea and administer a general anesthetic, which may exacerbate the BPD and create a situation in which extubation after surgery may be difficult to achieve. Either a spinal or epidural anesthetic may be used to provide surgical conditions with a regional block. For spinal anesthesia, an intrathecal injection of a hyperbaric solution of tetracaine (1 mg/kg) or isobaric bupivacaine (1 mg/kg) provides 1 to 2 hours of surgical anesthesia (see Chapter 41) The addition of epinephrine (10 µg) to the tetracaine solution increases the duration of anesthesia by 30 to 60 minutes.¹⁵⁴ Hypotension rarely occurs after spinal anesthesia in an infant.¹⁵⁷ For epidural anesthesia, 0.75 mL/kg of 0.375% bupivacaine with epinephrine is injected into the epidural space through the sacral hiatus.^155,156 Ultrasound can be used to map out the anatomy of the spinal cord and cauda equina, dura mater, and intrathecal space. During injection one can visualize the local anesthetic tracking up the epidural space to confirm its proper location (see also Chapters 41 and 42). Preterm infants are more sensitive to local anesthetic blockade than are children and adults; therefore surgical anesthesia may be achieved with reduced concentrations of local anesthetic. Another route to the epidural space is by the lumbar approach (L3-4 interspace).¹⁹ The use of the lumbar route allows a smaller dose of local anesthetic compared with the caudal route, thus decreasing the risk of toxicity, however, this route is technically more challenging.

Epidural anesthesia offers several advantages over spinal anesthesia in the awake infant, in that it is often easier to perform, and a catheter can be placed to repeatedly administer local anesthetic to extend the duration of the block (see Chapter 41). General anesthesia offers several advantages over regional anesthesia, including better operating conditions and greater ease in titrating anesthetic duration. An often cited advantage of regional anesthesia is the decreased incidence of postoperative apnea. However, the evidence in support of this notion is conflicting. When the incidence of postoperative apnea after spinal anesthesia, spinal anesthesia with IV ketamine, and general endotracheal inhalational anesthesia was compared,¹⁵⁸ no postoperative apnea was reported in the spinal anesthesia group, whereas an 89% incidence was noted in the spinal/ketamine group and a 31% incidence was noted in the general anesthesia group. Thus, if there is a benefit to regional anesthesia over general anesthesia, spinal anesthesia must be administered without supplemental sedatives, such as ketamine.

In contrast, Krane and colleagues compared general endotracheal inhalational anesthesia with spinal anesthesia and found no difference in the incidence of postoperative apnea.¹⁵⁹ In a study of more than 250 preterm infants receiving spinal anesthesia for inguinal hernia repair at a single institution experienced with this technique, spinal block was successfully placed on the first attempt in more than 90% of infants.¹⁶⁰ Despite a high rate of successful block placement, more than 20% of the infants required supplemental anesthesia at some time during the surgery. Postoperative apnea occurred in 4.9% of infants, and all infants who developed this complication had a preoperative history of apnea. Given the technical challenges of spinal block placement for practitioners who do not perform it routinely, failure rate even with successful block placement, and persistent risk of postoperative apnea, we commonly administer general anesthesia for hernia repair in the micropremie, except for those with severe BPD, in whom we use spinal anesthesia.

Eye Surgery for Retinopathy of Prematurity

ROP may be treated with cryotherapy, laser photocoagulation, or scleral buckling surgery and/or vitrectomy. Diode laser photocoagulation is typically performed at the bedside in the NICU for moderate ROP, whereas cryotherapy and scleral buckling surgery are performed in the OR. Cryotherapy involves applying a freezing probe under direct visualization to the avascular retina anterior to the fibrovascular ridge. Scleral buckling surgery and vitrectomy, performed for severe ROP with retinal detachment, are less frequently employed because early detection and treatment with laser photocoagulation prevents ROP progression to severe disease. Laser surgery has been shown to be as effective as cryotherapy for moderate ROP, and is most commonly used because the systemic side effects are significantly less, the ocular tissues are less traumatized, and this technique has a smaller incidence of late complications (see Chapter 32).

Laser photocoagulation may be performed under topical anesthesia alone, with IV sedation, or under general anesthesia. Cryotherapy and scleral buckling surgery require general anesthesia. For laser photocoagulation, the incidence of cardiorespiratory complications is greater with topical anesthesia alone than with topical anesthesia with sedation or general anesthesia.¹⁶¹ Factors that influence the selection of anesthetic technique include the infant’s medical condition, gestational age, and availability of pediatric anesthesia services. The vast majority (95%) of preterm infants who require laser photocoagulation or cryotherapy develop threshold disease between 32 and 42 weeks PCA. The usual age for scleral buckling or vitrectomy is much older, between 6 months and 1 year.

Laser photocoagulation takes 10 to 30 minutes to perform and often involves a series of treatments every few weeks. For laser photocoagulation, the anesthetic goals are to provide optic analgesia and to prevent eye and head movement. Many preterm infants younger than 32 weeks PCA are naturally inactive. With topical anesthesia alone they will remain motionless and do not require sedation or general anesthesia. As the infants mature beyond 32 weeks PCA, their activity naturally increases, along with the need for sedation and general anesthesia. Given the brevity and frequency of the procedure, we prefer a combination of topical anesthesia with an infusion of propofol and spontaneous ventilation through a natural airway. After applying the local anesthetic, we administer a series of IV propofol boluses (1 mg/kg) until there is no movement to tactile stimulation, and then begin a propofol infusion (150 µg/kg/min). We use an oral airway as necessary. We administer supplemental oxygen and monitor end-tidal CO₂ through a nasal cannula. Alternative sedative agents include IV midazolam (0.1 to 0.2 mg/kg), pentobarbital (2 mg/kg), or oral chloral hydrate (75 mg/kg). We employ standard monitors and conduct the procedure with the ophthalmologist, in a treatment room in the NICU; size-appropriate airway equipment should be immediately available should airway obstruction or apnea occur. Cryotherapy and scleral buckling surgery take more time to complete and require surgical preparation in the OR. As a result, our preferred technique is inhalational anesthesia with tracheal intubation and neuromuscular relaxant, along with topical anesthesia. After an inhalational induction with sevoflurane using standard monitors, we paralyze the child, intubate the trachea orally, and mechanically ventilate the lungs. Administration of caffeine reduces the incidence of postoperative apnea. At the end of surgery, we discontinue the inhaled agent, antagonize the neuromuscular blockade as appropriate, and extubate the trachea after the return of appropriate breathing and airway reflexes.

Anesthesia or Sedation for Radiologic Imaging

Very preterm infants frequently experience neurocognitive complications. With advances in the capabilities of MRI, and new therapies to improve neurologic outcome based on MRI findings, preterm infants are increasingly undergoing MRI evaluation. At many centers, virtually all preterm infants less than 30 weeks gestational age at birth undergo brain MRI during their hospitalization. In past years, head ultrasound and computed tomography were the modalities of choice to diagnose IVH and hydrocephalus. MRI is gradually replacing these modalities for this purpose because there is no exposure to radiation. Moreover, MRI is able to identify congenital lesions, vascular malformations, and ischemic injury, which the older modalities were less effective in detecting.

Although not painful, MRI requires absolute immobility for the duration of the scan, typically 45 to 60 minutes. Immobility may be achieved through simulated feeding, sedation, or anesthesia. In very preterm infants (less than 30 weeks PCA) and in good-tempered older preterm infants, swaddling the infant in warm blankets and applying sugar water (“sweet-ez”) to a pacifier often promotes natural sleep, enabling the scan to be obtained. In healthy preterm infants (“premie growers”) aged 30 to 70 weeks PCA, orally administered chloral hydrate (75 mg/kg) 30 to 60 minutes before the procedure provides sedation lasting 2 hours. We perform a history and physical examination and prescribe the sedation regimen, and specially trained sedation nurses in our department administer the regimen and monitor the infant during the scan, under our supervision. It should be noted that chloral hydrate has an extremely long half-life in this population, placing the infants at risk for late re-sedation and/or apnea.^162–164

In preterm infants with medical problems, we administer propofol anesthesia for the scan, titrated to effect similarly to that with laser photocoagulation for ROP. We monitor expired CO₂ through a nasal cannula, sometimes insert an oral airway, and less often use a laryngeal mask airway or an ETT. In critically ill preterm infants with an ETT and mechanical ventilation, we administer IV midazolam and fentanyl or inhaled sevoflurane for the scan. In anesthesia cases, monitors include electrocardiogram, blood pressure, pulse oximetry, and expired CO₂. IV pumps being used for administration of fluids or drugs must remain outside the MRI room, thus requiring extra-long tubing. All anesthesia equipment, the infant, and personnel must not contain ferrous materials. As with general anesthesia, monitoring for postoperative apnea for 24 hours after sedation is recommended (see Chapters 45 and 47).