Volume Control

In volume-control modes, the size of the breath is controlled by targeting a volume to be delivered. The ventilator senses air flow and calculates the volume as a product of flow and time. Most sophisticated volume ventilators measure and adjust the gas delivery on the basis of exhaled tidal volumes. When air leaks from the chest by routes other than the gas delivery circuit (e.g., a bronchopleural fistula or a leak around the endotracheal tube), volume delivery may be inaccurate. When compliance of the respiratory system is poor, a percentage of the tidal volume is lost in the circuit because of expansion of the ventilator tubing. The loss of tidal volume may be estimated using a compliance factor that is specific to the diameter of the ventilator tubing. The most sophisticated ventilators (the Servo 300i, Siemens Medical Solutions, Stockholm, Sweden) can adjust for this inaccuracy if that setting is selected. The most accurate measures of tidal volumes are obtained at the endotracheal tube using an in-line respiratory profile monitor (CO₂SMO Mainstream, Respironics). When the size of the breath is controlled by volume delivered, decreasing compliance in the lung will result in higher pressures achieved. An inspiratory pause gives the pressure in the circuit time to equilibrate between the proximal airway and the alveoli, and will help to determine if the alveoli are being distended by high pressures. In the setting of noncompliant, diseased lung, the duration of inhalation or inspiratory time will affect the pressures achieved in volume control modes.

Pressure Control

In pressure-control modes, the size of the breath delivered is determined by the target pressure and the inspiratory time. Pressure-control modes were developed for use in small infants before sophisticated methods existed to control delivery of very small tidal volumes. The volume delivered at a given airway pressure will depend on the compliance of the lungs. As the patient’s disease process resolves and compliance improves, tidal volumes may become large, placing the patient at risk for ventilator-induced lung injury (VILI). The very high initial flow rates necessary to achieve a given pressure for the duration of the inspiratory time can result in turbulence in a constricted airway, which results in decreased dynamic compliance and smaller tidal volumes delivered.

Pressure-Regulated Volume Control

Pressure-regulated volume control (PRVC) is a volume-controlled mode with a flow pattern similar to a pressure-controlled breath. A target tidal volume is set, and the ventilator applies constant pressure throughout inspiration to achieve that target volume. Volume delivered is measured on exhalation, and the pressure is adjusted in subsequent breaths to deliver the target tidal volume. Because exhalation volume is used to calculate the pressure needed for the next breath, PRVC may not be used when there is significant air leak outside the ventilatory circuit.

PRVC shares with pressure control the advantages of a larger area under the pressure-time curve, leading to higher mean airway pressure (MAP) and improvement in oxygenation. PRVC shares with volume control the safety of a controlled tidal volume that cannot increase as compliance improves or decrease as compliance worsens. This mode is particularly useful when dramatic improvement in compliance is expected, such as after the administration of surfactant. When PRVC is not available as a synchronized mode, significant inspiratory effort may result in breath stacking or extreme variability in tidal volumes delivered. For this reason, PRVC is usually used with very heavy sedation or neuromuscular blockade (or both), except with the Servo 300i, which can provide synchronized intermittent mandatory ventilation with PRVC.

Support Modes

Effort above a preset rate may or may not be supported. The most common form of support is pressure support. Pressure support has been shown to facilitate weaning, and it may also improve patient comfort. Pressure support is frequently used to recondition weakened respiratory muscles while still providing mandatory breaths to prevent atelectasis; however, data to support this approach do not exist (Turner and Arnold, 2007).

Ventilator-Induced Lung Injury

Ventilator-induced lung injury is caused by prolonged exposure to high Fio₂, by alveolar stretch from excessive tidal volumes (volutrauma), and by opening and closing of alveoli (shear stress trauma). Although hyperoxic inspired gas is essential for patients with lung injury, exposure to Fio₂ greater than 95% for 24 to 48 hours causes lung injury similar to ARDS and causes death within 48 to 72 hours due to oxygen toxicity (Adams et al., 2003).

Mechanical trauma from ventilation causes injury both through physical disruption of the architecture of the lung and through release of inflammatory mediators. Atelectatrauma, which is thought to occur when PEEP is set below the lower inflection point of the lung hysteresis curve, is particularly problematic in lung disease states associated with surfactant dysfunction, such as ARDS. Patients with heterogeneous lung disease are at increased risk for volutrauma. Transpulmonary pressure is not uniformly distributed to alveoli, and collapsed alveoli adjacent to inflated alveoli may be subject to extreme shearing forces. Mechanical trauma leads to release of inflammatory mediators, which may spill over to systemic circulation and play a role in the development of multiorgan failure (Plötz et al., 2004; Harris, 2005; Tremblay and Slutsky, 2006).

Use of Lung Hysteresis Curves

More sophisticated ventilators can calculate pressure-volume loops showing the difference in the pressure-volume relationship between the inspiratory and expiratory limbs of the respiratory cycle (Fig. 39-1). The sigmoidal shape of the hysteresis curve reflects differing compliance at different lung volumes. The lower inflection point represents the point during inspiration at which compliance improves. Below this point, the majority of alveoli are closed. The upper inflection point represents another change in compliance. Above a certain volume, more pressure is required to further distend alveoli, and the upper inflection point is thought to represent the beginning of a zone in which further pressure causes significant alveolar strain. The curve is steepest and nearly linear between these points, reflecting maximal compliance. Lung injury is thought to be minimized by keeping PEEP above the lower inflection point to maintain recruitment, and keeping tidal volumes below the upper inflection point by minimizing peak inspiratory pressure (Tremblay and Slutsky, 2006).

FIGURE 39-1 Pressure-versus-volume curve of the respiratory system. A, The inflation limb of the curve. B, The deflation limb of the curve. The point at which the curve touches the y-axis (zero pressure) is the functional residual capacity (FRC). The lower inflection point (LIP) represents the point during inspiration at which compliance improves. The upper inflection point (UIP) represents the point at which alveolar overdistention leads to worsening compliance. The curve is steepest between these points, reflecting the zone of best compliance. Lung injury is thought to be minimized by keeping positive end-expiratory pressure (PEEP) above the lower inflection point to maintain recruitment, and by keeping tidal volumes below the upper inflection point by minimizing peak inspiratory pressure.

(Reprinted with permission from Stawicki SP, Goyal M, Sarani B: High-frequency oscillatory ventilation [HFOV] and airway pressure release ventilation [APRV]: a practical guide, J Intensive Care Med 24:215, 2009.)

High-Frequency Oscillatory Ventilation

Despite conventional ventilation strategies to reduce VILI, about one third of patients with ARDS die, and many of the survivors experience a significant decrement in function. The object of high-frequency oscillatory ventilation (HFOV) is to accomplish oxygenation and ventilation with less injury to the lung. HFOV uses laminar flow at high MAPs to maintain the lung open, and it uses an oscillating piston to create pressure displacement above and below the MAP to deliver very small tidal volumes in the range of the volume of the patient’s anatomic dead space (Stawicki et al., 2009) (Fig. 39-2). The result, ideally, is ventilation only on the steep and linear part of the hysteresis curve. Compliant alveoli, which might receive an excessively large share of the delivered tidal volume in conventional ventilation, are protected from volutrauma. With this strategy, noncompliant alveoli—those with time constants long enough to preclude opening with conventional ventilation—may be recruited. Recruitment of new lung units may improve V/Q matching. A crossover trial comparing HFOV with conventional ventilation in children with severe lung disease showed that HFOV was associated with improved oxygenation and, despite higher MAPs, a reduced need for supplemental oxygen at 30 days (Arnold et al., 1994).

FIGURE 39-2 High-frequency oscillatory ventilation (HFOV). Both graphs depict pressure versus time in HFOV. The graph on the left depicts the ventilator cycle over only a fraction of a second, to illustrate the key variables that are controlled in HFOV. The graph on the right shows the pressure-versus-time curve of HFOV, with conventional ventilation superimposed for comparison. The y-axes are different, reflecting airway pressures in HFOV on the left and pressures in conventional ventilation on the right. The exhalation pressure in HFOV is actually negative, making expiration an active process.

(Reprinted with permission from Stawicki SP, Goyal M, Sarani B: High-frequency oscillatory ventilation [HFOV] and airway pressure release ventilation [APRV]: a practical guide, J Intensive Care Med 24:215, 2009.)

In HFOV, the variables controlled on the ventilator are frequency of respiration, (measured in cycles per second, or Hertz), amplitude of ventilation (power), and MAP. Frequency is measured in cycles per second. A frequency of 10 is 600 “breaths” per second, and a frequency of 3 is 180 breaths per second. A lower frequency functionally creates a longer inspiratory time and therefore a larger breath. The frequency is initially set at 10 for infants, and somewhat lower for older children. Amplitude, also referred to as power, is measured in centimeters of water. There is no standard starting pressure. The amplitude is initially set to produce a visible wiggling motion of the patient’s body to the level of the lower abdomen or groin. MAP is initially set to 5 cm H₂O greater than the last MAP on conventional ventilation (Stawicki et al., 2009). A chest radiograph performed within an hour or two after initiating HFOV should confirm an appropriate MAP setting by showing approximately nine visible ribs from the top of the thorax to the diaphragm.

Oxygenation can be improved by increasing Fio₂, MAP, or bias flow. Bias flow is an auxiliary flow of gas that crosses the oscillating gas flow to provide fresh gases and clear CO₂. Increasing MAP to the point where alveolar distention causes collapse of perialveolar vessels may result in hypotension or hypoxemia that may be responsive to administration of intravascular volume. Lung recruitment often allows a decrease in MAP within the first 24 hours. When the MAP can be weaned to near the last MAP recorded from conventional ventilation, conversion back to conventional ventilation should be considered.

Ventilation can be improved by increasing the amplitude and decreasing the frequency. As the frequency of breaths decreases, a higher percentage of the set amplitude will be delivered. This is like increasing the tidal volume in a conventional ventilator; decreasing frequency is a way to increase ventilation, thereby decreasing Paco₂. Frequencies less than 3 (180 cycles per minute) may increase the risk for volutrauma (Stawicki et al., 2009).

Patients on HFOV are usually treated with sedation and neuromuscular blockade. Because they are subject to a constant distending pressure, which may impede venous return, it is important to be attentive to maintaining a good intravascular volume and right ventricular filling pressure. Many patients require intravascular volume loading at the initiation of HFOV. Although nursing staff may be reluctant to suction patients on HFOV, secretions in the endotracheal tube can dramatically attenuate amplitude and should be removed. Finally, the appearance of a patient on HFOV, with the loud noise of the machine’s diaphragm and the absence of tidal movement of the chest, can be disconcerting to patient’s families. It is best to inform them of what to expect.

For a more complete treatment of the mechanics of HFOV, see Stawicki and colleagues (2009).

Airway Pressure-Release Ventilation

Airway pressure-release ventilation (APRV) cycles between a high- (P_high) and a low-pressure level (P_low) in a high-flow CPAP circuit (Fig. 39-3). The continuous high flow allows spontaneous breathing throughout the ventilatory cycle, and it may reduce the need for heavy sedation and neuromuscular blockade. Spontaneous breathing is necessary to realize the benefits of this mode of ventilation (Stawicki et al., 2009).

FIGURE 39-3 Airway pressure-release ventilation (APRV). In the pressure-versus-time curve for APRV, spontaneous breaths are present throughout the respiratory cycle.

(Reprinted with permission from Stawicki SP, Goyal M, Sarani B: High-frequency oscillatory ventilation [HFOV] and airway pressure release ventilation [APRV]: a practical guide, J Intensive Care Med 24:215, 2009.)

Most of the respiratory cycle is spent at P_high, and pressure is briefly released to P_low several times per minute. Ideally, time spent at P_high will maximize recruitment and, by placing the lung on the steep part of the hysteresis curve, make spontaneous breathing easier. The time spent at P_low allows ventilation, but it is believed to be brief enough to prevent atelectasis of alveoli with long time constants (Stawicki et al., 2009).

Initially, P_high is set at the plateau pressure measured from an inspiratory pause on conventional ventilation and adjusted to allow weaning of Fio₂. P_low is set at zero to maximize the velocity of exhalation. The time (T) of T_high is initially set between 4 to 6 seconds and adjusted to maintain adequate oxygenation and ventilation. T_low is set at 0.4 to 0.5 seconds and adjusted to maintain a tidal volume of about 5 mL/kg, or such that expiration ends when the flow reaches 50% to 75% of the measured maximal expiratory flow. T_low is the critical variable that affects bulk air flow and thus ventilation. Oxygenation is increased by increasing Fio₂ or by increasing P_high or T_high, both of which will increase MAP. Ventilation is increased by encouraging spontaneous respiration or increasing T_low (Stawicki et al., 2009).

Although it was described in 1987, APRV is relatively new in pediatrics and has not been thoroughly studied. Case reports of APRV in pediatric patients have shown improvement in oxygenation with the ventilation strategy (Krishnan and Morrison, 2007; Kuruma et al., 2008). With a crossover design, the one randomized controlled trial of APRV in children demonstrated comparable ventilation at lower peak inspiratory and plateau pressures (Schultz et al., 2001). The potential advantages of decreasing mechanical ventilatory pressures and allowing spontaneous breathing and a reduced need for sedation and neuromuscular blockade in patients with severe lung disease make it worth considering despite the paucity of data. APRV may be contraindicated in patients with severe obstruction to exhalation, such as status asthmaticus.

Helium-Oxygen Mixtures (Heliox)

Helium is a low-density, biologically inert gas that can be safely mixed with oxygen. When air flow through a tube is turbulent, the flow rate depends on the density of the gas. With high helium concentrations, heliox can flow more rapidly through narrowed airways than nitrogen-oxygen mixtures. Additionally, carbon dioxide diffuses through helium faster than through air. The Fio₂ required by the patient may limit the use of heliox, as minimal benefit is believed to occur from heliox concentrations of less than 60%. When used with a spontaneously breathing patient, efforts should be made to prevent entrainment of room air with a well-fitting nonrebreather mask. Heliox can be delivered through the ventilator but may interfere with tidal volume measurements. This difficulty can be overcome with in-line monitors of respiratory mechanics (Myers, 2006).

Heliox may improve the delivery of nebulized medications. It improves dyspnea and peak expiratory flow in spontaneously breathing asthmatics, and it may limit peak inspiratory pressure in ventilated patients with status asthmaticus (Gupta and Cheifetz, 2005; Kissoon et al., 2008). Heliox improves respiratory distress associated with upper airway obstruction and improves the work of breathing in patients with bronchiolitis (Gupta and Cheifetz, 2005). No adverse effects of heliox have been reported. Despite a widespread belief that early application of heliox can prevent the need for intubation in select patients, there are no studies of heliox directed at patient outcomes.

Prone Positioning

In a supine patient, the dependent lung is underventilated and at risk for collapse, and the antidependent lung is aerated and at risk for overdistention. The mechanism by which prone positioning can improve oxygenation is not completely clear, but it may improve V/Q matching by improving the aeration of the dependent lung (Ryu et al., 2008). Prone positioning can also improve lung mechanics by decreasing thoracoabdominal compliance. In adult patients, a protocol of prone positioning for 7 hr/day improved oxygenation but did not alter ventilator-free days or mortality (Gattinoni et al., 2001). In pediatric patients, a protocol of prone positioning for 20 hr/day improved efficiency of oxygenation but did not improve ventilator-free days or organ failure–free days (Curley et al., 2005).

Nitric Oxide

Inhaled nitric oxide (iNO) has been a mainstay of treatment for both primary and secondary pulmonary hypertension of the newborn since its introduction. In these patients, it causes a rapid and sustained improvement in efficiency of oxygenation by preferentially vasodilating ventilated lung units (Konduri and Kim, 2009). Because it is rapidly inactivated by hemoglobin, the effect of iNO is limited to the pulmonary vasculature (Adhikari et al., 2007).

Because the pathophysiology of ARDS and acute lung injury (ALI) involves V/Q mismatching, there has been interest in using iNO for patients outside the neonatal age range. In adult studies, iNO improved oxygenation (i.e., the ratio of pulmonary arterial oxygen pressure [Pao₂] to Fio₂, and the oxygenation index) but did not improve mortality. There has been some concern in adult patients that iNO may be associated with an increase in the need for new renal replacement therapy (Adhikari et al., 2007). In a randomized trial in children, iNO improved oxygenation without harmful side effects from methemoglobinemia at dosages up to 20 parts per million. The study was not powered to detect a difference in mortality (Tang et al., 1998).

Nitric oxide is commonly used at a level of 20 ppm. Very high concentrations of NO may paradoxically worsen V/Q matching by diffusing into nonventilated areas and improving the perfusion there. High concentrations also carry a significantly increased risk for methemoglobinemia. Methemoglobin should be monitored daily while therapy with iNO is used. If methemoglobin becomes elevated, it is usually sufficient to decrease the concentration of iNO. For very high methemoglobin levels, a scavenger, such as methylene blue, may be used. Exposure to NO for even a brief period can sensitize the pulmonary vasculature to its effects. Rapid weaning to low levels (5 ppm) is usually well tolerated, but thereafter, iNO should be weaned slowly to prevent rebound vasoconstriction (Adhikari et al., 2007).

Surfactant

Surfactant deficiency has been described in ARDS and ALI, particularly in the acute phase of the illness. The best-studied application of surfactant is in meconium aspiration syndrome, where significant clinical improvement is reported after surfactant instillation (Willson et al., 2008). Surfactant instillation in infants mechanically ventilated for respiratory syncytial virus disease leads to improvement in oxygenation and decrease in duration of mechanical ventilation (Tibby et al., 2000; Luchetti et al., 2002).

Trials of endotracheal instillation or aerosolization of surfactant in adults have shown little or no benefit. In children, treatment with calfactant, a natural preparation relatively rich in surfactant specific protein B, has been associated with improved oxygenation and mortality, with particular benefit seen in patients younger than 12 months (Willson et al., 2005). Although this trial generated significant interest in the use of calfactant for pediatric ARDS and ALI, an independent analysis of the data noted that the placebo arm of the trial had more immunocompromised patients, and that statistical significance was lost when controlling for immune status (Czaja, 2007). Surfactant therapy for ARDS/ALI in patients outside of the neonatal age range remains controversial.

Surfactant is administered by direct endotracheal instillation through an endotracheal tube or bronchoscope at a dosage of 100 mg/kg body weight. The total volume of the drug preparation is significant. Complications of surfactant instillation include volume-responsive hypotension and transient hypoxia. Most authors cite pulmonary hemorrhage as a relative contraindication (Willson et al., 2008). Surfactant is probably best used early in the course of mechanical ventilation. Administration of surfactant in patients with severe hypoxia can provoke severe desaturation. In patients with severe hypoxia, surfactant may be safely given while the patient is supported with extracorporeal therapies.

Asthma

Children who are ventilated for status asthmaticus are among the most difficult ventilated patients to manage. Resistance to exhalation may cause intrinsic PEEP, which can profoundly decrease the available inspiratory volume. In severe asthmatics, exhalation is an active process; under neuromuscular blockade, the loss of patient participation in exhalation almost always results in significant worsening of hypercarbia. The most important strategy for mechanical ventilation of the patient with status asthmaticus is to prevent the need for positive pressure ventilation with aggressive bronchodilator and steroid therapy. Helium-oxygen mixtures and intravenous magnesium can be useful adjuncts to these therapies.

NPPV, if applied early in patients with status asthmaticus, can take advantage of patient’s respiratory efforts and prevent the muscle fatigue that leads to the need for intubation (Stather and Stewart, 2005). There is evidence that NPPV prevents hospitalization in patients presenting to the emergency department with status asthmaticus. Mental status changes, anxiety, exhaustion, or refractory hypoxemia preclude the use of NPPV, and it is these clinical changes rather than arterial blood gas analysis results that should guide the decision to intubate.

Once the patient is intubated, the primary strategy is to limit gas trapping, which causes intrinsic PEEP and forces the lung to operate on the less compliant portion of the hysteresis curve. Gas trapping is evident in flow-versus-time graphs, where inspiration is seen to start before flow returns to zero, or in graphs of end-tidal CO₂ versus time, where CO₂ fails to reach a plateau before the next inspiration begins (Stather and Stewart, 2005).

Significant controversy exists as to the best method for ventilating asthmatics. There are advocates for both pressure- and volume-control ventilation. Because of the high air-flow rates required to achieve targeted pressure in pressure-control ventilation, this mode may cause more turbulence in the airways of ventilated asthmatics. Volume-control modes may require higher set pressures to deliver adequate tidal volumes; however, because much of the pressure is lost in overcoming proximal airway resistance, alveolar hyperinflation may not occur. Alveolar pressure can be measured with an inspiratory pause pressure; complications are rare when this pressure does not exceed 30 cm H₂O (Stather and Stewart, 2005).

Intrinsic PEEP can be measured with an expiratory pause pressure in a muscle-relaxed patient. Because complete exhalation is the goal, some authors recommend minimizing PEEP with an initial setting from 0 to 5 cm H₂O. Others, acting on the theory that PEEP will stent airways open and ameliorate gas trapping, recommend setting PEEP at or slightly higher than the measured intrinsic PEEP. If a higher-PEEP strategy is used, intrinsic PEEP should be measured frequently, and the ventilator settings should be adjusted accordingly (Stather and Stewart, 2005).

There is general agreement that exhalation time should be as long as possible. Low respiratory rates and short inspiratory times give maximum exhalation time, but often at the expense of minute ventilation. Permissive hypercapnia (see Acute Respiratory Distress Syndrome and Acute Lung Injury, next) has been successfully used in the ventilation of asthmatics (Stather and Stewart, 2005). With permissive hypercapnia, patient ventilatory dyssynchrony is common. Neuromuscular blockade with adequate sedation is often necessary until severe bronchospasm resolves, usually within 24 hours. Weaning and extubation should be prompt, because spontaneous respiration is favorable for asthmatics. There are no generally agreed on criteria for extubating asthmatics, except that it should be performed as soon as possible. Negative inspiratory pressure with active exhalation is more effective than positive pressure in obstructive lung disease. Blood gas and ventilatory parameters that suggest the patient with asthma is improving include a normalizing partial pressure of CO₂ and pH, and a peak inspiratory pressure that begins to approach the inspiratory pause pressure. The difference between peak inspiratory pressure and inspiratory pause pressure correlates with airway resistance.

Acute Respiratory Distress Syndrome and Acute Lung Injury

The clinicopathologic picture of ARDS and ALI is diffuse alveolar damage, and four clinical parameters are involved: acute onset, severe arterial hypoxemia resistant to supplemental oxygen, bilateral infiltrates, and absence of left atrial hypertension. A Pao₂/Fio₂ ratio of less than 200 mm Hg defines ARDS, and a ratio of less than 300 mm Hg defines ALI (Randolph et al., 2003).

The most common cause of ARDS and ALI is lower respiratory tract infection, but it can be precipitated by a variety of systemic inflammatory conditions, including infection outside the respiratory tract, pancreatitis, and tissue necrosis. Overall mortality for children ranges between 8% and 15%, which is lower than that for adults (Randolph et al., 2003). Patients with bone marrow transplantation may have a significantly higher mortality. Because of the low mortality in children, duration of mechanical ventilation or number of ventilator-free days are both considered acceptable endpoints for studies of ARDS and ALI.

Although not a predictor of mortality in adults with ARDS, multiple single-center and multicenter studies of the epidemiology of ARDS and ALI in pediatric patients confirm that the ratio of the alveolar oxygen pressure (Pao₂) to Fio₂ at the onset of ARDS or ALI is an independent predictor of mortality (Flori et al., 2005; Erickson et al., 2007). Mortality rates for children with ARDS or ALI from studies in the past 10 years range from 8% to 35% (Willson et al., 2008; Randolph et al., 2003). A recent prospective multicenter observational study of 320 children with ARDS or ALI found that mortality in patients with a Pao₂/Fio₂ ratio of less than 200 is nearly twice the mortality in those with a ratio between 200 to 300 (26% versus 13%) (Erickson, 2007). The Pao₂/Fio₂ ratio also predicted the number of ventilator-free days (Flori et al., 2005). The severity of hypoxia is also sometimes measured by the oxygenation index, defined as (MAP × Fio₂)/Pao₂ (Randolph, 2009).

Treatment of ARDS and ALI

Control of a persistent trigger of the inflammatory cascade is critical to eliminating the stimulus for ongoing lung injury. Because inflammation with capillary leak is part of the pathophysiology of ARDS, patients may benefit from a restrictive fluid strategy (Wiedemann et al., 2006). Fluid restriction should be initiated only after a patient with shock has been adequately resuscitated. Treatment of hypoproteinemia with albumin and furosemide may also be beneficial (Randolph, 2009).

Intensive care unit (ICU) stays for patients with ARDS or ALI may be prolonged, so is important to be attentive to other organ systems and to the prevention of iatrogenic injury. Strategies to avoid ICU-related pneumonia, critical care myopathy, central line infection, deep venous thrombosis, and gastritis, will be discussed later (see Prophylaxis).

Mechanical ventilation strategies for ARDS and ALI are designed to minimize VILI while protecting the patient from profound hypoxia. A Pao₂ target of 55 to 80 mm Hg (i.e., an arterial oxygen saturation as measured by pulse oximetry [SpO₂] of 88% to 95%) is commonly recommended for adult patients but may be too low for children, as the effect of prolonged marginal oxygenation on the developing brain is unknown. Most authors recommend an SpO₂ target of greater than 90% for children (Randolph, 2009).

In patients with ARDS and ALI, parenchymal lung damage occurs heterogeneously. The relatively normal aerated lung, which may be a relatively small proportion of the total lung volume, is at risk for overdistention. In 2000, the Acute Respiratory Distress Syndrome Network (ARDSNet) trial, in adults, a strategy aimed at reducing overdistention decreased mortality. The authors recommended reducing plateau pressure to 30 cm H₂O or less by targeting tidal volumes to 6 mL/kg or less. These data have not been replicated in children. Adoption of the target values from the adult study is controversial, particularly for younger children, as their more compliant chest wall and smaller airways may predispose them to atelectasis.

Alveolar collapse is possible with a low tidal volume strategy. Recruitment maneuvers with such a strategy include application of a sustained increase in airway pressure (i.e., a pressure of 40 mm Hg for 40 seconds) with a goal of opening the collapsed lung. After recruitment, PEEP is optimized to maintain the open lung (Randolph, 2009).

Permissive Hypercapnia

If the mode of ventilation that maintains ideal oxygenation and minimizes VILI is not capable of adequate CO₂ exchange, a lower pH and a higher Pco₂ may be tolerated. Patients in whom hypercapnia is being tolerated to minimize lung stretch will exhibit increased air hunger and anxiety and will require increased sedation. Hypercapnia is accompanied by acidosis in the acute phase, until renal compensation can buffer the pH. Hypercapnic acidosis (HCA) may benefit patients with ARDS or ALI apart from its association with ventilator strategies designed to decrease lung stretch. HCA increases cardiac output and organ blood flow (Rogovik and Goldman, 2008). It has antiinflammatory effects and attenuates free radical production. It may thus function to attenuate further lung injury from mechanical ventilation and from evolving infection.

HCA may have deleterious effects in select populations. Decreases in neutrophil chemotaxis and in the bactericidal activity of neutrophils and macrophages seen with HCA may be harmful, particularly in patients with sepsis (Curley et al., 2010). HCA increases pulmonary vascular resistance, and therefore it may not be appropriate for patients who are anticipated to have reactive pulmonary vasculature, particularly those in the neonatal age range (Curley et al., 2010). The increase in organ blood flow with HCA can cause increased intracranial pressure (ICP) in at-risk patients. Higher levels of CO₂ are not believed to damage the brain, but rigorous long-term outcome studies in children have not been done (Curley et al., 2010).

Extracorporeal Membrane Oxygenation

Despite all the innovative therapies, deterioration of lung function in patients with acute respiratory failure may lead to a requirement for 100% oxygen and high ventilatory pressures. In some of these patients, extracorporeal membrane oxygenation (ECMO) can ensure adequate oxygenation and ventilation without high Fio₂ and without potentially injurious airway pressures, allowing time for the lungs to heal. ECMO has been a particularly effective treatment for neonates with persistent pulmonary hypertension of the newborn: the use of ECMO reduced mortality in this group from between 80% and 85% to about 20% (Roy et al., 2000). Although older trials had suggested that ECMO for acute respiratory failure in adults was futile, the Extracorporeal Life Support Organization registry reports a survival rate of 53% for the small number of adult patients treated with ECMO for respiratory failure (Zapol et al., 1979; Conrad et al., 2005). A recent randomized trial of ECMO for adults with acute respiratory failure found the therapy effective in terms of improving survival without disability, and in terms of cost effectiveness (Peek et al., 2009). The use of ECMO was associated with an improved survival in pediatric patients with respiratory failure (Green et al., 1996). In pediatric patients, the mortality reported by the Extracorporeal Life Support Organization registry was 56% (Conrad et al., 2005). A randomized controlled trial of ECMO in pediatric patients was attempted but failed because of an overall decrease in mortality, probably resulting from lung-protective ventilation strategies.

The general indication for ECMO is reversible lung disease that prevents adequate oxygenation with conventional therapy, or shock refractory to medical treatment in a patient who is believed to have reversible disease (see ECMO for Circulatory Support of the Patient with Shock, later).

It is difficult to precisely identify the indications for ECMO for the patient with respiratory failure. The issue is clearer in neonates, who may be expected to have a relatively rapid resolution of pulmonary hypertension and a good outcome with extracorporeal therapy. Before the advent of high-frequency oscillating ventilation and nitric oxide, an oxygenation index of 40 was the criterion for considering ECMO. A recent review of 174 newborns with respiratory failure suggested that using a lower oxygenation index (the authors suggested 33.2) as the criterion for ECMO resulted in a lower mortality rate and a lower rate of chronic lung disease (Bayrakci et al., 2007). As mechanical ventilation becomes more sophisticated, it becomes more difficult to identify its moment of failure. With continuous innovation, it is also difficult to collect historical data for criteria that would identify patients in whom mortality with conventional ventilation exceeds the expected mortality with ECMO.

ECMO should be considered in a patient with high airway pressures and high Fio₂ who may be expected to require a long course of mechanical ventilation, or when the clinician believes that the probability of chronic lung disease induced by mechanical ventilation outweighs the risk of ECMO. The best outcomes are seen in patients with single-organ dysfunction and viral pneumonia. Mortality is highest in patients with trauma, immune suppression, or multiple-organ dysfunction syndrome (MODS) at the time of ECMO initiation. Most authors consider prolonged ventilatory course (>10 to 14 days) with high pressures and high Fio₂ to contraindicate ECMO, because significant VILI has already occurred (Frenckner and Radell, 2008). Survival to hospital discharge is less likely in patients who have been ventilated longer than 10 days (Nehra et al., 2009). Absolute contraindications to ECMO include conditions that make cannulation impossible, including severe prematurity or small size (e.g., infants <34 weeks old and weighing <2 kg), any condition that would preclude systemic heparinization (e.g., major coagulopathy, active hemorrhage, or recent intracranial hemorrhage), or conditions amenable to surgical repair (e.g., total anomalous pulmonary venous return and lethal anomalies).

Once the need for ECMO is identified, the route of cannulation must be determined. Venoarterial ECMO, usually performed through the jugular vein and carotid artery, is used to provide both respiratory and cardiac support. In patients with respiratory failure who lack a significant component of cardiac insufficiency, venovenous ECMO may be preferable. Venovenous ECMO eliminates the risk for arterial embolic events and preserves the carotid artery, but, because of the inevitable recirculation, it may not allow complete oxygenation.

When ECMO is initiated, the ventilator is adjusted to settings that allow the lung to rest. Generally, this means reducing the Fio₂ to 21% and the airway pressures to the minimal settings that allow some lung inflation. In patients older than neonates, it can be very difficult to reopen lungs that have completely collapsed on ECMO.

The pulmonary hypertension that leads to neonates requiring ECMO resolves in a few days, but the lung damage that leads to pediatric patients requiring ECMO is much slower to heal. For pediatric patients, the mean time on ECMO is almost 2 weeks (Frenckner and Radell, 2008; Nehra et al., 2009). A recently published 8-year experience with ECMO from Massachusetts General Hospital described an average duration of ECMO for respiratory failure of 274 hours (11.4 days), with a range of 1 to 1154 hours (48 days) (Nehra et al., 2009). The duration of a course of ECMO should be considered on a case-by-case basis. In general, it is reasonable to continue ECMO support for a pediatric patient with respiratory failure until there is clear evidence that pulmonary disease is irreversible, or until complications preclude survival.

Every day on ECMO increases the risk for complications. Bleeding occurs in up to 16% of pediatric patients on ECMO, and it is most commonly seen at surgical sites, at cannula sites, and in the GI tract. Complications related to the ECMO circuit (e.g., oxygenator failure, cannula dysfunction, tubing rupture, pump malfunction) occur for 14% of pediatric patients (Frenckner and Radell, 2008). During prolonged ECMO runs, attention to the support and preservation of extrapulmonary organ systems becomes paramount. Patients who may require ECMO should be sent to a center with expertise in the procedure before the severity of lung disease prevents safe transport.

For a thorough review of management of the patient on ECMO, see Frenckner and Radell (2008). For the most complete resource on ECMO management, see Van Meurs and colleagues (2005).

Fluid Refractory Shock

Vasoactive agents are an important adjunct to the stabilization of children in shock, but they should not be considered a replacement for adequate fluid resuscitation. Current ACCM guidelines recommend beginning with an inotrope (dopamine, up to 10 mcg/kg per minute, or lower-dosage epinephrine, beginning at 0.05 mcg/kg per minute). Mortality increases with delay in initiation of inotropes. Inotropes may be given through peripheral venous access (with close monitoring of the access site) or through an intraosseous line (Carcillo et al., 1991). In patients who require vasoactive medications, central venous access should be established. Warm shock should be treated with norepinephrine delivered via a central line (Carcillo et al., 1991).

Catecholamine-Resistant Shock

When shock is not reversed by the addition of centrally delivered inotropes or vasopressors, adrenal insufficiency should be considered (Carcillo et al., 1991). The diagnosis of relative adrenal insufficiency in patients with septic shock is controversial. Hypothalamic pituitary adrenal axis suppression can occur in patients with sepsis by a variety of mechanisms, as a result of dysfunction at any point along the hypothalamic pituitary adrenal axis. A landmark study by Annane and colleagues (2002) found an improvement in mortality rates when septic patients with relative adrenal insufficiency were treated with steroid replacement. This study found a high incidence of relative adrenal insufficiency, which was probably related to a high rate of etomidate exposure in these patients, leading the author to recommend abandoning the use of etomidate (Annane, 2005). A subsequent multicenter randomized controlled trial failed to find a mortality benefit, although corticosteroid replacement did hasten reversal of shock in patients when shock could be reversed (Sprung et al., 2008). Exogenous corticosteroids have been shown to improve the vascular response to exogenous catecholamines in patients with shock (Oppert et al., 2005). The definition of an adequate circulating cortisol level is a subject of debate and probably depends on the severity of the illness. The most widely accepted definition in adults is a serum cortisol level of less than 15 mcg/dL, or an increase of less than 9 mcg/dL after ACTH stimulation (Annane et al., 2000). Because the dosage used in the ACTH stimulation test is supraphysiologic, some question its usefulness in identifying patients who could benefit from steroid replacement. Current recommendations for adult sepsis recommend use of low-dosage steroids (<300 mg/day of hydrocortisone) for shock that is resistant to fluid resuscitation and catecholamines without use of ACTH stimulation test (Dellinger et al., 2008; Marik et al., 2008). Because the P450 cytochrome system, which is responsible for steroidogenesis, matures to an adult level of functioning by about 1 year of age, infants and young children are at higher risk for adrenal insufficiency (Bartelink et al., 2006; Masumoto et al., 2008; Fernandez and Watterberg, 2009).

Adrenal insufficiency should be suspected in patients with fluid- and catecholamine-resistant shock who have risk factors for adrenal insufficiency, including a history of systemic steroid use, with a rapidly evolving purpuric rash, a history of panhypopituitarism, or exposure to etomidate (Brierley et al., 2009). The ACCM guidelines recommend hydrocortisone replacement for children with risk factors for adrenal insufficiency and catecholamine-resistant shock and for those with shock and inadequate response to the corticotropin stimulation test (peak cortisol, <18 mcg/dL after stimulation) (Brierley et al., 2009). A cortisol level should be drawn before the first dose of hydrocortisone. It is not always possible to wait for results of a stimulation test before initiating steroid replacement therapy. The recommended dosage for hydrocortisone ranges from 2 mg/kg per day (stress dosage) to 50 mg/kg per day (shock dosage) (Brierley et al., 2009). The dosage may be titrated to resolution of shock. Results of serum cortisol, when available, may lead to discontinuation of replacement hydrocortisone. The patient should be weaned from steroid therapy as soon as hemodynamics allow, because the impact of exogenous stress-dosage steroids on sensitive elements of the immune system is not fully understood. Prospective randomized controlled trials in children have not been done.

Increasing Oxygen Delivery and Normalizing Perfusion

The 2007 clinical practice parameters for hemodynamic support of pediatric and neonatal septic shock from the ACCM recommend use of clinical goals to direct shock resuscitation in the first hour. Shock resuscitation should be continued until therapeutic endpoints are reached (Box 39-1). Adult studies have shown improved outcomes when therapies for shock are titrated to maximize oxygen delivery, using venous oxyhemoglobin saturation (Svo₂) as a surrogate (Rivers et al., 2001). After the first hour of resuscitation in children, shock management should be directed to generating an Svo₂ of greater than 70% and, when pulmonary artery (PA) catheters are used, a cardiac index of 3.3 to 6.0 L/min per square meter (Brierley et al., 2009). A low Svo₂ may be treated with increasing inotropes or with blood transfusion, aiming for a hemoglobin level of about 10 mg/dL.

In adults, the use of PA catheters has not been shown to improve mortality, but comparable studies have not been done in children. PA catheters have been shown to identify incorrect assessment of hemodynamic status made on the basis of clinical parameters in children (Ceneviva et al., 1998). PA catheters may be useful in children when therapy directed by central venous pressure and Svo₂ is inadequate to reverse shock (Carcillo and Fields, 2002).

Children with cold shock and with poor perfusion often benefit from a reduction in cardiac afterload. Phosphodiesterase inhibitors such as milrinone are useful adjuncts to adrenergic agonists in these patients. Many of these children already have maximal adrenergic stimulation of the myocardium (whether exogenous or endogenous) and can benefit from the alternative pathway to increased inotropy provided by milrinone. As phosphodiesterase inhibitors are initiated, many patients will require further volume expansion. Loading dosages of phosphodiesterase inhibitors (e.g., milrinone) should not be used in pediatric patients with shock. Continuous intravascular monitoring is necessary when afterload reducers are used for patients with shock.

Vasopressin has been used in catecholamine refractory shock, particularly in the setting of decreased vascular tone. Children with severe shock often have high levels of endogenous catecholamines and some degree of catecholamine resistance. This, coupled with the fact that catecholamines have some unfavorable effects on immune function and catabolism, have led investigators to consider other biochemical pathways to reverse shock. Vasopressin is a powerful vasoconstrictor, but at low dosages it has organ-specific vasodilatory effects (Holmes et al., 2003). Vasopressin deficiency has been described in vasodilatory shock states in both adults and children (Landry et al., 1997; Lodha et al., 2006). Studies in adults have shown that vasopressin improves blood pressure and urine output and allows lowering of the dosage of norepinephrine; however, a recent large randomized controlled trial of low-dosage vasopressin infusion found a good safety profile but no impact on mortality when compared with norepinephrine alone (Russell et al., 2008).

Vasopressin and its synthetic analogue, terlipressin, have been used in children in two ways: as a rescue therapy for severe shock, and as a catecholamine-sparing hormone replacement. Case series of the use of vasopressin in patients with refractory shock have described the use of dosages in the range of 0.00001 to 0.08 units/kg per minute (Matok et al., 2005; Choong and Kissoon, 2008; Yildizdas et al., 2008). When vasopressin is used as a hormone replacement, it is not titrated to clinical effect but used as a low-dosage continuous infusion. The only pediatric randomized controlled trial so far used a dosage of 0.0005 units/kg per minute, as in adult trials, and found an increase in MAP but no differences in time to hemodynamic stability or other clinical outcomes (Choong et al., 2009).

Reassessment of Fluid Balance and Hemodynamic Profile

Serial reassessments of patients admitted with shock are critical to good outcomes, and require continuous and meticulous attention to the maintenance of optimal preload, contractility, and afterload. Shock in neonates and children is more variable than in adults. Children who present in warm shock with an increased cardiac output and low systemic vascular resistance often progress to cold shock with myocardial dysfunction within 24 hours. Shock that is refractory to fluid resuscitation and vasoactive agents should prompt review of volume status. A high percentage of patients who have persistent shock after volume resuscitation have an underfilled ventricle, seen by echocardiography (Jardin et al., 1999). It should also raise the possibility of alternative diagnoses, including acquired cardiac disease, obstructive etiologies, toxins, occult blood loss, and inborn errors of metabolism.

Other Causes of Shock

Once shock resuscitation is initiated, the clinician should begin to consider the causes. Septic shock is the most common etiology of shock in the PICU. In patients with persistent shock, mechanical obstruction to flow in the form of pericardial effusion or elevated intraabdominal pressure should be considered. Other causes of shock, including occult hemorrhage, toxicologic causes, and acquired heart disease, should be addressed.

Causes of shock in the neonatal age range include congenital heart disease, particularly ductal-dependent lesions. An infusion of prostaglandin E₂ should be started immediately in neonates with shock until a ductal-dependent congenital heart lesion can be ruled out. Prostaglandin may cause apnea at higher dosages. Very young infants still transitioning from fetal to neonatal circulation are also at high risk for pulmonary hypertension and right-sided heart failure. Acidosis from shock and hypoxia can trigger this pathophysiology. Therapies that reduce pulmonary arterial pressure, including supplemental oxygen, sedation, maintenance of an alkalotic pH, and nitric oxide, are commonly needed for neonates with fluid-refractory shock and may be started empirically if necessary (Carcillo and Fields, 2002). Particular attention should be paid to maintenance of adequate temperature and glucose level in neonates with shock.

Early antibiotic administration is a high priority in patients presenting with septic shock. In adults with septic shock, a delay in antibiotic therapy is associated with worse survival, with mortality rates increasing by 7% for every 30 minutes that passes without delivery of appropriate antibiotic therapy (Kumar et al., 2006). Cultures are extremely important, but antibiotics should not be delayed if it is difficult to secure appropriate samples. In particular, lumbar puncture should be deferred until the patient is more stable and an evaluation of the coagulation system can be performed.

ECMO for Circulatory Support of the Patient with Shock

Shock that is refractory to medical interventions may be treated with extracorporeal support. ECMO for pediatric circulatory support was used in 29% of all cases in 2004, but most of these were cardiac patients requiring perioperative support (Conrad et al., 2005). Shock related to acute fulminant myocarditis has been successfully treated with extracorporeal support, with survival rates approaching 80% for patients on ECMO (Duncan et al., 2001). ECMO may be a bridge to myocardial recovery or to heart transplantation in these patients. It may be more effective in the support of shock characterized by inadequate cardiac output than of shock with a prominent component of vasoplegia or failure of oxygen extraction.

The use of ECMO for other causes of shock is uncommon and poorly studied. Historically, sepsis was considered a contraindication to ECMO because of concerns that the circuit would be come irreversibly contaminated, but today, sepsis causing pulmonary hypertension is a standard indication for ECMO in neonates. Randomized controlled trials of ECMO for refractory septic shock are not possible, because the condition is rare and the options for treatment are few. A recent single-center report of 45 pediatric patients treated with ECMO for refractory circulatory failure due to sepsis, collected over 18 years, is the largest study published to date. These authors reported a 47% survival to hospital discharge despite the fact that 91% of the patients had three or more organ failures (Maclaren et al., 2007). Estimates of mortality without ECMO in case series of patients in which ECMO was used for this indication support the idea that ECMO as a treatment for septic shock significantly improves mortality rates (Bartlett, 2007). The reported average durations of ECMO for sepsis-related circulatory failure are 137 hours and 76 hours in case series of 9 and 12 patients, respectively (Beca and Butt, 1994; Goldman et al., 1997).

ECMO should be used for refractory shock when the clinician believes that other options are exhausted and the patient is unlikely to survive without it. In the treatment of septic shock, ECMO provides the treating team time to identify and treat the infectious agent, and it eliminates the problems of high dosages of vasoactive drugs and the toxicities associated with treatment of patients with ALI related to systemic inflammation.

Management of the Patient after Cardiac Arrest

Protocols for the management of pediatric cardiac arrest have been developed by the American Heart Association and are covered in Chapter 38, Cardiopulmonary Resuscitation. The focus here is on the care of children after the return of spontaneous circulation.

The most important treatment for pediatric cardiac arrest is prevention. Because pediatric cardiac arrest is not usually sudden in onset, a window of opportunity exists in the period before it occurs. Medical emergency teams designed to respond to patients in danger of decompensation have been shown to decrease hospital mortality rates and the frequency of more serious codes (Topjian et al., 2009). In addition to early recognition of impending cardiac arrest, extracorporeal cardiopulmonary resuscitation (ECPR) has been used in an attempt to improve these outcomes. Intact survival from isolated respiratory arrest is the rule, but outcomes from in-hospital pediatric cardiac arrest remain poor. Return of spontaneous circulation is established in one half to two thirds of cases of in-hospital arrest, but less than 18% of patients survive with good neurologic outcome (Nadkarni et al., 2006). Survival rates and neurologic outcomes for out-of-hospital cardiac arrest are much poorer (Young et al., 2004). Cold-water submersion for cardiac arrest, even for relatively long duration, may be associated with a much higher rate of intact survival.

Extracorporeal Cardiopulmonary Resuscitation

Multiple physicians have examined the application of extracorporeal therapy as a rescue therapy during cardiac arrest before the return of spontaneous circulation. Survival to hospital discharge in these case series is surprisingly high—35%—despite a relatively long median duration of cardiopulmonary resuscitation (CPR) before the establishment of flow on ECMO (mean duration of 50 minutes) (Morris et al., 2004). With ECPR, survival to hospital discharge does not correlate with duration of CPR before ECMO. Neurologic outcomes have been encouraging: good neurologic outcomes are reported in greater than 75% of patients (Huang et al., 2008; Prodhan et al., 2009). It is important to select patients who have a reversible disease process for use of ECPR.

Management of Hypoxic-Ischemic Brain Injury after Resuscitation

Strategies aimed at preservation of brain function after cardiac arrest have been reported in adults and are under investigation in children. There is no single measure to decrease the impact of cardiopulmonary arrest on the brain.

Postischemic cytotoxicity occurs in animal models, so eliminating pathologic states known to increase active cerebral metabolism, such as seizure or hyperthermia, may be beneficial. The majority of cell injury may occur during the initial hypoxic ischemic event. However, cell death from apoptosis and as a result of reperfusion continues after the initial insult. After the event, an initial period of hyperemia is followed by a period of delayed postischemic hypoperfusion. Therapeutic strategies aimed at controlling this delayed cell death include maintenance of normoglycemia; avoidance of oxygen toxicity; control of states characterized by increased cerebral metabolic rate, such as seizures and hyperthermia; and the application of therapeutic hypothermia.

Controlled hypothermia is associated with improved mortality rates and improved functional outcome after cardiac arrest in adults and in neonates with birth asphyxia (Bernard et al., 2002; Hypothermia after Cardiac Arrest Study Group, 2002; Gluckman et al., 2005; Shankaran et al., 2005). It is difficult to extrapolate these findings directly to children, because the circumstances surrounding cardiac arrest in children are different from those in adults. Ventricular fibrillation is the most common cause of cardiac arrest in adults, but cardiac arrest in children is predominately caused by asphyxia. Asphyxial arrest occurs after a period of hypoxia, and the total duration of cerebral insult (hypoxia plus anoxia) may be very long. Ischemic injury may be seen in brain cells after relatively short periods of interrupted circulation if they are preceded by a period of hypoxia (Vaagenes et al., 1997).

Clinical trials of therapeutic hypothermia after cardiac arrest in children are ongoing. Although there are not yet randomized controlled data supporting controlled hypothermia for pediatric patients after cardiac arrest, it is being used in the United States “consistently” by 9% of the pediatric intensivists surveyed and “sometimes” by 38% (Haque et al., 2006). Target temperatures vary widely, with most using temperatures of 32° to 35° for 12 to 24 hours (Haque et al., 2006). It is relatively labor intensive to rapidly achieve and maintain hypothermia. Neuromuscular blockade may be necessary to control shivering. Slow and controlled rewarming (0.5° C every 2 hours) is an important part of this therapy. Vigilance during the rewarming phase is necessary to avoid hyperthermia, which is known to exacerbate ischemic brain injury, but also to allow careful monitoring of the blood pressure and electrolyte disturbances that are common with rewarming (Wang et al., 2009; Fink et al., 2010a, 2010b). In a retrospective analysis of children after cardiac arrest, there was no increase in the number of hypothermia-associated adverse events in patients who received this therapy (Doherty et al., 2009).

Hypoglycemia has a synergistic, deleterious effect when coupled with neonatal asphyxia (Lubchenco and Bard, 1971). Hyperglycemia has been shown to be detrimental after near-drowning and traumatic brain injury, and, in general, glucose-containing solutions are not used in the first 48 hours after brain injury in older children (Ashwal et al., 1990; Chiaretti et al., 2002). Both hypotension and hypertension during reperfusion have been shown to have deleterious effects in animal models (Miller and Myers, 1972; Bleyaert et al., 1980).

Oxygen toxicity is posited to play a role in postischemic reperfusion injury. The brain is susceptible to oxygen free-radical damage (Siesjö et al., 1989). On the basis of this, there has been some interest in using room air to resuscitate newborns, particularly those with preserved circulation (Ten and Matsiukevich, 2009). In older patients, avoidance of hyperoxia may be beneficial.

Seizure activity increases cerebral metabolic rate and may worsen ischemic injury. Subclinical seizure activity is common after cardiac arrest, particularly in patients treated with therapeutic hypothermia, because of the requirement for neuromuscular blockade. Prophylactic use of antiepileptic drugs in patients who require paralysis is reasonable (Abend et al., 2009).

Prediction of Outcome after Cardiac Arrest

Attempts to use variables such as duration of cardiac arrest, initial pH, or early neurologic examination to predict neurologic outcome after cardiac arrest in children have been misleading. Electroencephalography can provide some prognostic information. The burst suppression pattern seen on the electroencephalogram after cardiac arrest is sensitive and specific for poor outcome (Wijdicks et al., 2006). No single test has been validated to predict outcome with a high sensitivity and specificity. High-quality CPR, and well-established ventilation and oxygenation in the period before the return of spontaneous circulation or the establishment of ECMO flow, may be the best predictors of outcome (Topjian et al., 2009).

Criteria for Initiating Renal Replacement Therapy

There are no nationally published guidelines for the initiation of RRT, and there is wide variation among practitioners as to its timing of initiation and discontinuation. Absolute indications for renal replacement therapies include life-threatening electrolyte abnormalities and anuria. However, RRT is more commonly used for the relative indication of fluid overload with inadequate urine output (Gibney et al., 2008). The guidelines used at the Children’s Hospital of Pittsburgh (University of Pittsburgh Medical Center) are shown in Box 39-2. There is increasing evidence that fluid overload contributes to mortality, particularly in children. Observational studies suggest that increasing severity of fluid overload before or at initiation of RRT is an independent predictor of mortality after controlling for severity of illness. RRT should be considered when accumulated fluid overload reaches 10% (Ricci and Ronco, 2008; Sutherland et al., 2010).

Practical Aspects of Renal Replacement Therapy

Continuous renal replacement therapy (CRRT) is the preferred mode of renal replacement therapy in ICU patients with hemodynamic instability. The advantages of CRRT over intermittent hemodialysis are that it allows correction of metabolic acidosis, provides temperature control, and provides constant fluid hemostasis with better avoidance of intracerebral fluid shifts. Its main drawback is the need for anticoagulation. Although CRRT shows no survival benefit over intermittent therapy, it may predict renal recovery in survivors (Uchino et al., 2007).

CRRT can be accomplished with ultrafiltration alone (i.e., continuous venovenous hemofiltration [CVVH]) or with dialysis. With CVVH, both water and high-molecular-weight substances pass across the membrane because of a pressure gradient that creates convective flow. Convection, or passive movement of solute across a membrane along with water, allows phosphates and molecules such as urea and creatinine to clear at similar rates. Profound hypophosphatemia can easily develop, and patients on CVVH require phosphate replacement. Larger molecules such as heparin, insulin, and vancomycin, cleared in only negligible amounts during dialysis, are cleared more efficiently by CVVH.

With dialysis, blood flows along one side of a semipermeable membrane as a crystalloid solution is pumped along the other side of the membrane, countercurrent to the blood flow. Through the process of diffusion, molecules cross the membrane. The larger the molecule, the more slowly it moves across the membrane. Urea, a small molecule, is cleared efficiently, whereas creatinine, a larger molecule, is cleared less well. Phosphate ions have very low rates of clearance across the membrane.

Clinical Changes When Initiating CRRT

Clinicians should be alert to the risk for hypotension when CRRT is initiated. Hypotension may occur as a result of inadequate intravascular volume, hypocalcemia, or bradykinin release syndrome, particularly when patients are acidotic at the time of CRRT initiation. Once CRRT is begun, therapies aimed at preventing fluid overload should be discontinued. Diuretics should be decreased or stopped. Nutrition should be increased if it had been limited by fluid restriction. The amount of protein in parenteral nutrition should be increased to 1.5 to 4 mg/kg per day, because hemofiltration results in significant amino acid loss across the membrane (Maxvold et al., 2000). A low blood urea nitrogen level in a patient on RRT reflects inadequate nitrogen balance. A patient on CRRT requires additional potassium and phosphorus to replace that lost to the circuit. When CRRT is discontinued or interrupted, diuretics, fluid rate, TPN prescription, and electrolyte infusions should be returned to normal. The removal of medications across the CRRT membrane is difficult to predict. Consultation with a pharmacist is recommended with initiation of CRRT.

High-Clearance CRRT for Treatment of Sepsis

It is conventional to speak of the dosage of CRRT, meaning the quantity of blood purified per unit time. Higher-clearance CRRT (at least 35 mL/kg per hour versus 20 mL/kg per hour) is associated with improved survival in adults, with the greatest improvement seen in septic patients (Ronco et al., 2000). There has been some interest in using extremely high clearance CRRT to remove cytokines and modulate inflammation. This approach has shown some promise in patients with ARDS after bone marrow transplantation (DiCarlo et al., 2003). The pore size in most filters used for hemofiltration and dialysis admit passage of proteins whose size is in the range of 30 to 40 kDa, which is more than adequate for removal of interleukins 1, 6, 8, and 10, and tumor necrosis factor. Studies of high-clearance CRRT in septic patients have not shown significant and sustained decreases in cytokine levels, and high-clearance CRRT has not been conclusively shown to improve outcome (Joannidis, 2009).

Contraindications to CRRT

There are no absolute contraindications for the use of CRRT. Patients at risk for life-threatening bleeding with systemic heparinization may be put on CRRT with a citrate anticoagulation protocol. Citrate is added to the CRRT circuit, generating a level of hypocalcemia in the circuit sufficient to prevent clotting. Calcium must be infused centrally to replace the calcium lost in the circuit. Most of the citrate is removed in the circuit, and the liver metabolizes what remains. When citrate is used for anticoagulation, patients with liver failure who are unable to metabolize citrate are at risk for citrate lock, manifested by an increased ratio of total to ionized calcium (Morgera et al., 2004). Risks and benefits should be weighed carefully when considering the use of citrate CRRT in patients with liver failure.

Enteral Feeding

Enteral feeding is usually withheld until cardiovascular stability is established, because generalized hypoperfusion and vasoactive agents limit intestinal blood flow. However, early enteral feeding may preserve gut mucosal integrity, may decrease septic episodes, and may decrease the incidence of ventilator-associated pneumonia (Jeejeebhoy, 2001; Kompan et al., 2004). Aspiration is thought to be a risk factor for ventilator-associated pneumonia (VAP), but enteral nutrition decreased the risk for VAP in adults. Similar studies have not been done in children. The frequent interruptions to enteral feeding (for many reasons, including perceived intolerance, procedures, diagnostic tests, medication administration) may result in inadequate nutrition delivery (de Neef et al., 2008). Attention should be given to calories delivered, not just calories prescribed. Feeding protocols may limit the impact of these factors on the delivery of adequate nutrition (Meyer et al., 2009). When complete enteral nutrition is impossible, low-dosage enteral nutrition in combination with parenteral nutrition is probably preferable to parenteral nutrition alone (Heidegger et al., 2008).

Transpyloric feeding may help to avoid unnecessary feeding interruptions, particularly in patients with gastroparesis or those who are at increased risk for aspiration, but it does not prevent aspiration of gastric contents (Meert et al., 2004). Because infants and children are at higher risk for malplacement of enteral feeding tubes, tube placement should be confirmed radiographically before use. Enteral formulas exist for special situations such as renal failure, general malabsorption, fat malabsorption, and pancreatitis. Although it is generally accepted that enteral nutrition is superior to parenteral nutrition, studies in pediatric patients are lacking.

Parenteral Nutrition

Total parenteral nutrition (TPN) is indicated for patients with contraindications to enteral feeding. TPN is associated with a number of complications, including intestinal mucosal atrophy with increased bacterial translocation, metabolic abnormalities including hyperglycema and hypertriglyceridemia, and immune dysfunction and infection. With long-term therapy, hepatic steatosis, cholestasis, and irreversible liver failure can occur. When there is risk for liver injury, TPN is often cycled to mimic the postabsorptive state (Kelly, 2006). Timing of initiation of parenteral nutrition is controversial. A recent review suggested starting it when enteral nutrition could not be used for 3 to 5 days (Skillman and Wischmeyer, 2008). In patients at high risk for malnutrition, it may be appropriate to initiate parenteral nutrition sooner.

In TPN, carbohydrates provide 60% to 75% of nonprotein calories. Carbohydrate concentrations are increased gradually to avoid hyperglycemia. The glucose infusion rate is calculated according as follows:

The glucose infusion is typically started at 5 to 6 mg/kg per minute and increased by 0.5 to 1 mg/kg per minute per day for preterm infants, or 2 to 4 mg/kg per minute per day for older children. For patients with sepsis, the ACCM guidelines recommend at least 8 mg/kg per minute for neonates, 5 mg/kg per minute in children, and 2 mg/kg per minute in adolescents (Brierley et al., 2009). Glucose control may be important in critically ill patients (see later). Some of the negative effects of parenteral nutrition seen in studies done in the past two decades may be related to carbohydrate overfeeding and poor glycemic control.

Provision of adequate protein intake prevents negative nitrogen balance and skeletal muscle wasting and supports growth and wound healing. To prevent protein from being used as a calorie source and to preserve its availability for growth and repair, adequate nonprotein calories must be provided (see Table 39-2). In renal failure, it may be necessary to decrease the amount of protein delivered. When CRRT is initiated, protein delivery should increase to compensate for protein lost across the membrane.

Fats serve as an energy source and a source of essential fatty acids. About 30% to 40% of TPN calories are delivered as fat, providing both an energy source and a source of essential fatty acids. At least 3% to 4% of calories should be given as fat to prevent free fatty acid deficiency. If too much fat is given, the lipoprotein lipase system is overloaded and lipemia occurs. TPN fats are delivered as a 20% emulsion, which is low in osmolarity and thus safe for peripheral veins.

It is routine to provide a pediatric multivitamin in TPN. Other minerals, such as selenium, iodide zinc, and iron, may not be provided by standard TPN solutions but are available as additives.

Intensive Insulin Therapy in Critically Ill Patients

Hyperglycemia is common in critically ill patients. In children, hyperglycemia is associated with organ dysfunction and poor outcomes (Kyle et al., 2010). Maintenance of normal fasting blood glucose values (80 to 100 mg/dL) conferred a survival advantage over maintenance blood glucose values (180 to 200 mg/dL) in critically ill adult surgical patients and was associated with a shorter ICU and hospital stay, a shorter period of mechanical ventilation, and less renal injury in adult medical patients (van den Berghe et al., 2001, 2006). In a recently published trial of tight glucose control using age-stratified glucose targets in critically ill children, tight control was associated with improvement in morbidity and mortality over a glucose target of 180 to 214 mg/dL, despite episodes of hypoglycemia occurring in 25% of the treatment group. More than two thirds of the children in this study were postoperative cardiac surgery patients; this may hamper the study’s generalizability to medical patients (Rouette et al., 2010).

The appropriate target for glycemic control in critically ill pediatric patients remains to be determined. Adoption of a tight glucose control strategy requires frequent monitoring and familiarity with glycemic control protocols at all levels of care. The long-term impact of hypoglycemia in the PICU is not known. These issues have prevented the widespread adoption of tight glycemic control in critically ill children.

Physiology of Traumatic Brain Injury

In the injured brain, there is an area of primary injury that cannot be ameliorated, a penumbra of brain tissue that is at risk for further injury, and normal brain distant from the site of initial injury. The goal of critical care after TBI is to keep that penumbra as small as possible and to protect the normal brain from further injury. Only part of the injury sustained in TBI occurs with the initial injury. The most important causes of secondary injury include hypoxia, hypotension, hyperglycemia and hyperthermia, seizures, and intracranial hypertension. The mechanisms for secondary brain injury, including excitotoxicity, have been reviewed by Kochanek and coworkers (2008).

Initial Management of the Brain-Injured Patient

Early management of airway, breathing, and circulation should take into account the devastating effect of even transient hypoxia and the impact of hypercarbia on ICP. Prevention of hypoxia by intubation is strongly associated with improved outcomes (Franschman et al., 2009; Zebrack et al., 2009). The initial management of TBI is presented in Box 39-3. The examiner should always be sensitive to signs of intracranial hypertension or impending herniation, including pupillary dilation, Cushing’s triad, and extensor posturing. The Glasgow Coma Scale has been adapted for use in children and should be calculated for all patients with TBI on arrival and in subsequent examinations (Table 39-3). Brain injuries evolve over time, and the initial computed tomography scan might not reflect the current state of brain edema or intracranial hemorrhage. A sustained change in ICP or in the neurologic examination should prompt repeat imaging.

TABLE 39-3 Glasgow Coma Scale for Children and Adults

Management of Intracranial Hypertension

Intracranial hypertension may impair cerebral blood flow and is associated with poor outcome. It results from osmolar shifts, cellular edema, vasogenic edema (blood-brain barrier breakdown), mass lesions, and intracranial hemorrhage. Protocols exist for the management of intracranial hypertension, but they should not be seen as a surrogate for cooperative management with a neurosurgical team (Adelson et al., 2003). Therapies aimed at controlling intracranial pressure are applied in a stepwise fashion, beginning with the least invasive measures (Box 39-5).