Chapter 44. Liberation from Mechanical Ventilation

• Patients are candidates for liberation from mechanical ventilation when gas exchange or circulatory disturbances that precipitated respiratory failure have been reversed.
• More than half of all critically ill patients can be liberated successfully from mechanical ventilation after a brief trial of spontaneous breathing on the first day that reversal of precipitating factors is recognized. Gradual reduction of mechanical support, termed weaning, frequently is unnecessary and can prolong the duration of mechanical ventilation.
• Once a patient has been liberated from the ventilator, extubation should follow if mechanisms of airway maintenance (e.g., cough, gag, and swallow) are sufficient to protect the airway from secretions. Whether to extubate is a decision that follows successful liberation from the ventilator.
• In patients who fail their first trial of spontaneous breathing, attention should turn to defining and treating the pathophysiologic processes underlying failure.
• Weaning regimens that use ventilator modes with the goal of improving respiratory muscle endurance have not been proved to expedite liberation, but data from animal models suggest that “exercise" may be beneficial. Approaches attempting to exercise the respiratory muscles should not substitute for daily interrogation of readiness for spontaneous breathing.
• One weaning regimen, the gradual reduction of intermittent mandatory breaths, prolongs patients' time on mechanical ventilation.
• Liberation from mechanical ventilation is achieved most expeditiously if patients with a stable circulation (not on pressors or with evolving myocardial infarction) and adequate oxygenation are given a trial of spontaneous breathing (T piece or pressure support ≤7 cm H2O) each day. Patients remain on ventilators unnecessarily when clinicians do not put this simple plan in place.
• Patients who have had most correctable factors addressed and remain marginal with regard to ventilatory capacity in most circumstances should undergo a trial extubation rather than remain intubated for protracted periods. Noninvasive positive-pressure ventilation is extremely useful in these patients to transition them to fully spontaneous breathing following extubation.

Respiratory failure occurs when the lungs and respiratory pump fail to exchange oxygen and carbon dioxide adequately (see Chap. 31). Hypoxemic (type 1) respiratory failure usually results from flooding or collapse of the distal airspaces leading to intrapulmonary shunt and inadequate arterial oxygenation despite generous concentrations of inspired oxygen (see Chap. 38). Hypercapnic (type 2) respiratory failure results from inability to sustain sufficient alveolar ventilation to eliminate the CO2 produced from aerobic metabolism. Perioperative respiratory failure, a special case of types 1 and 2, results when postoperative pain and recumbency result in atelectasis and hypoxemia or when medications to alleviate pain reduce respiratory drive, leading to hypercapnia. Shock-related respiratory failure is another special case in which the underperfused respiratory muscles are unable to compensate for the acidosis resulting from inadequate global tissue perfusion. Mechanical ventilation substitutes for the respiratory pump until these disturbances have been reversed adequately to allow resumption of spontaneous breathing and gas exchange.

Institution of positive-pressure ventilation can be lifesaving for the disorders just described but is also associated with many complications (Table 44-1), and if this form of life support is extended beyond the point at which it is absolutely necessary, it may result in unnecessary morbidity and mortality for the patient. Most studies have demonstrated that earlier withdrawal of mechanical ventilatory support, when feasible, is associated with better outcomes for the patient. We will discuss the cardiopulmonary effects of positive-pressure ventilation and then outline principles and approaches to the withdrawal of mechanical ventilation in a way to achieve this milestone of critical care management at the earliest possible time and in the safest possible fashion.

Pulmonary Effects

Positive-pressure ventilation (PPV) is the most common mode of mechanical ventilation used in critically ill patients. PPV can deliver an inspiratory volume or pressure to the lungs and can provide positive end-expiratory pressure (PEEP). PEEP reduces the likelihood of end-expiratory alveolar collapse, thus preserving functional residual capacity.1,2 When used properly, PPV and PEEP can achieve and maintain alveolar recruitment and improve oxygenation. Excessive PPV or PEEP levels also can have adverse effects on gas exchange by reducing cardiac output or by increasing physiologic dead space.2,3

The discontinuation of mechanical ventilation increases the propensity for atelectasis, especially in patients with respiratory muscle weakness, restrictive physiology (e.g., obesity), or respiratory depression. In patients with lung injury, surfactant depletion and ultrastructural lung changes increase the likelihood of alveolar collapse (see Chap. 38). Thus cessation of PPV or PEEP may lead to atelectasis and hypoxemia (Table 44-2). In addition, the transition to spontaneous ventilation frequently is associated with increased venous return, which can precipitate pulmonary edema and hypoxemia. Finally, for patients with unrepaired imbalance between respiratory system load and neuromuscular competence, discontinuation of mechanical ventilation will precipitate the symptoms and signs of recurrent ventilatory failure (Fig. 44-1).

Cardiovascular Effects

The principal circulatory effects of PPV and PEEP are to reduce biventricular preload4–9and left ventricular afterload.4,10–13 PPV and PEEP also may increase right ventricular afterload by increasing lung volumes.14 When patients begin to breathe spontaneously, intrathoracic pressures that were positive on inspiration during PPV and PEEP become negative. Accordingly, the transition from PPV and PEEP to unassisted breathing causes acute increases in venous return and left ventricular afterload. In patients with normal hearts, these circulatory changes usually are well tolerated. However, in patients with left ventricular dysfunction, augmentation of venous return and increased left ventricular afterload increase left heart transmural pressures substantially. In addition, increased cardiac loads during weaning may precipitate ischemia in patients with coronary artery disease. Thus the transition from PPV to spontaneous breathing is accompanied by multiple events that can contribute to left ventricular failure and cardiogenic pulmonary edema (Fig. 44-2).

In order to minimize the duration of ventilator dependence, the clinician must:

1. Identify the pathogenesis of respiratory failure in each patient and institute appropriate treatment based on this pathophysiology.

2. Prevent iatrogenic complications associated with mechanical ventilation.

3. Detect when the patient is ready to breathe without the machine.

Since mechanical ventilation has numerous risks, including infection and barotrauma15–17 (see Table 44-1), it is appropriate to work aggressively to repair the “broken” patient and determine each day whether the patient still requires the ventilator. In this section we delineate the physiologic changes that occur during and the impact of underlying diseases on the transition to spontaneous breathing. We highlight methods used to expedite readying the patient for liberation and to identify patients who are appropriate candidates for liberation from mechanical ventilation. We also describe an approach to patients who do not succeed rapidly at being liberated from mechanical ventilation.

Item 1: Fix the Patient

Although it may seem intuitive that an organized systematic approach aimed at remedying the pathogenesis of disease should expedite liberation from mechanical ventilation, only recently has a study examined the efficacy of such an approach. Smyrnios and colleagues18 examined the utility of a combined protocol to identify and repair pathophysiology that binds patients to ventilators and recognize readiness to breathe spontaneously. This approach resulted in a substantial reduction in ventilator days and costs and expedited patient throughput. However, it is not possible to determine the degree to which the specific systematic approach of fixing the patient versus the liberation protocol contributed to the observed improvements. While from this study or the available literature it is not possible to determine the relative importance of each of these arms of treatment of respiratory failure, it is certain that both are crucial. Accordingly, from the very first day a patient requires intubation, it is worthwhile to define the mechanisms causing the need for initiation of mechanical ventilation. This can be approached through the following questions:

1. Did the patient need the ventilator, an artificial airway, or both? A significant number of patients are intubated for airway incompetence in the absence of significant hypoxemia or causes of hypercapneic respiratory failure. Disorders leading to an incompetent airway include primary neurologic diseases that attenuate airway protective reflexes or primary mechanical airway problems such as airway obstruction or insufficient cough to expectorate secretions.

2. If the patient needed the ventilator, was the mechanism of respiratory failure acute hypoxemia, hypercapnia (ventilatory failure), or a combination of these abnormalities? This is usually apparent from the clinical presentation combined with data obtained shortly after intubation. Patients whose initial presentation is poor oxygenation despite high concentrations of inspired oxygen (i.e., 100% facemask) most often have shunt physiology with atelectasis and/or alveolar flooding with blood, pus, or edema. Patients with acute hypercapnia can be separated into two categories: those with diminished drive to breathe (e.g., narcotic/sedative overdose) manifesting as a relatively comfortable patient with diminished level of consciousness and no respiratory distress or those who are uncomfortable and tachypneic, suggesting an imbalance between respiratory muscle strength and endurance and the load on the respiratory system. Some patients present with elements of both elevated work of breathing and hypoxemia, such as a patient with pneumonia with both a large intrapulmonary shunt producing hypoxemia and the consolidated lung increasing lung elastance and the work of breathing.

Hypotheses regarding pathogenesis can be confirmed shortly after intubation by evaluation of the chest radiograph and by using the mechanical ventilator as a pulmonary function machine. Lobar atelectasis usually is seen as an abnormal opacity, but subsegmental atelectasis can cause a shunt that is not obvious on a chest radiograph. Atelectasis often improves with PPV unless caused by airway obstruction. Patients with hypoxemia requiring high FiO2 and PEEP levels and lung opacities generally have alveolar flooding processes. Other clinical data are then used to confirm the mechanism of flooding. Appropriate history, purulent endotracheal secretions, and leukocytosis point to pneumonia. A dropping hematocrit with or without bloody endotracheal secretions suggests alveolar hemorrhage. Four-quadrant, relatively homogeneous airspace filling on chest radiograph suggests pulmonary edema (in the absence of hemorrhage), and attention turns to determining whether left ventricular dysfunction and/or hypervolemia are responsible or if capillary leak has resulted in hypoxemia (i.e., acute lung injury, acute respiratory distress syndrome). Clinicians should define the pathophysiology and treat underlying causes as appropriate.

In patients with hypercapnic respiratory failure, attention turns to examination of the elements of neuromuscular capacity and respiratory load. Although primary respiratory “pump” failure occurs occasionally (requiring one to consider a series of neurologic disorders, including CNS depression, polio or ALS, Guillain-Barré syndrome, phrenic nerve dysfunction, myasthenia gravis, and muscle dysfunction due to electrolyte deficiencies, medications, or sepsis; see Fig. 44-1), most commonly hypercapnic failure results from overloading of the neuromuscular apparatus. The three respiratory muscle loads are

1. Flow-resistive pressure cost of breathing. This is a function of the magnitude of flow and the resistance of the conducting airways, where Pr = flow × resistance. When the patient is on the ventilator and flow is delivered in a constant (square-wave) regime, Pr = (peak airway pressure -plateau airway pressure)/flow. If measured with a constant flow of 60 L/min (1 L/s), Pr = peak – plateau pressure yields resistance in centimeters of water per liter per second. These are the units used in the pulmonary function laboratory; since a 6- to 8-mm endotracheal tube and ventilator tubing contribute 5 to 10 cm H2O/L per second and normal airway resistance is less than 5 cm H2O/L per second, total resistance that is greater than 10 to 15 cm H2O/L per second after the patient is intubated is abnormal. Removal of airway secretions and administration of bronchodilators and, if required, corticosteroids are used to treat elevated airway resistance related to airway disease. Left-sided heart dysfunction also may contribute to increased airway resistance (cardiac asthma) and should be considered in patients with known or suspected cardiac disease.

2. Elastic pressure cost of breathing. This refers to the pressure required to expand the lung and chest wall from functional residual capacity to a given tidal volume; Pel = Ers × Vt, where Ers = elastance of the respiratory system and Vt = tidal volume. The elastic (plateau) pressure can be measured on the ventilator with an inspiratory pause of 0.5 to 1.0 second. The inverse of the elastance—the compliance—can be computed as Vt/plateau pressure – PEEP, normal being 50 mL/cm H2O. If compliance is decreased (elastance increased), one must determine whether the lung or chest wall is responsible. Chest wall noncompliance, which arises from conditions such as abdominal distention, obesity, and kyphoscoliosis, is usually readily apparent by clinical examination. Lung noncompliance usually is associated with lung infiltrates, large pleural effusions, or dynamic hyperinflation (due to increased airway resistance and/or hyperventilation).

3. Minute volume. This is not a mechanical load of breathing, but insofar as a higher minute volume requires more resistive and elastic pressure work of breathing (more breaths per minute), a higher minute volume loads the neuromuscular apparatus. Since minute volume equals alveolar ventilation plus dead space, any process that increases these will increase minute volume requirements. Alveolar ventilation can be increased by excess CO2 production, such as occurs in sepsis/systemic inflammatory responses and overfeeding, or by inefficient elimination (high dead space), such as occurs in late-stage acute respiratory distress syndrome (ARDS), chronic obstructive pulmonary disease (COPD), and pulmonary embolism.

After stabilizing patients on a ventilator, the three loads are measured easily, and if they are elevated, they likely contributed to respiratory muscle failure and hypercapnia. If they are completely normal, one reconsiders the possibility of either diminished drive to breathe or a cause of neuromuscular weakness that includes the respiratory muscles.

Hypercapnic respiratory failure occurs clinically in one of three ways: acute, chronic, and acute on chronic. In acute hypercapnic respiratory failure, a sudden mismatch of respiratory load to neuromuscular competence places the patient in acute respiratory failure over minutes to hours. There is an abrupt increase in PaCO2 with no metabolic correction and a blood pH consistent with an acute respiratory acidosis. In chronic hypercapnic respiratory failure, the mismatch of respiratory load to neuromuscular competence has been long-standing and usually gradual in evolution—over days to weeks or even years. PaCO2 elevation exists, but metabolic correction is near complete. Typically, pH is 7.30 to 7.36. These patients are particularly vulnerable to acute increments of load or decrements of capacity, termed acute-on-chronic hypercapnic respiratory failure. The respiratory muscles are unable to respond to the new insult, and PaCO2 rises too fast for immediate renal compensation. pH is usually less than 7.30, PaCO2is much greater than 40 mm Hg, and bicarbonate is elevated (>24 mEq/L), the “footprint of chronicity.” In such cases, clinicians should seek to understand the pathogenesis of the chronic component and to define which loads and/or decrements of capacity have contributed to the acute component.

Item 2: Prevent Iatrogenic Complications of Mechanical Ventilation

Although not emphasized in most discussions of weaning from mechanical ventilation, strategies that avoid further lung injury during mechanical support are extremely important to returning the patient to spontaneous breathing. Ventilator-induced lung injury (VILI) (see Chap. 37) refers to a number of mechanisms by which lung injury can occur in patients with acute lung injury and ARDS.17 Patients with severe airflow obstruction are at risk for dynamic hyperinflation and adverse consequences such as hypotension and diminished venous return20 (see Chap. 40). In both cases, ventilation with plateau airway pressures of less than 30 to 35 cm H2O appears to be associated with improved outcomes. Some patients with excessive chest wall stiffness (e.g., severe obesity) may require higher plateau airway pressures to maintain alveolar recruitment, but we seldom use tidal volumes that yield plateau airway pressures in excess of 35 cm H2O. A common misconception is that the peak airway pressure is the primary marker for risk of lung injury—indeed, this is an alarm parameter usually followed during mechanical ventilation. Very high peak airway pressures are well tolerated by the patient if plateau (i.e., end-inspiratory) alveolar pressures are maintained at less than 35 cm H2O, a circumstance encountered commonly in patients with status asthmaticus undergoing mechanical ventilation (see Chap. 40).

Aspiration should be prevented by maintaining the head of the bed of all ventilated patients at 30 degrees unless contraindicated.21 Ventilated patients usually are sedentary and therefore at risk of deep venous thrombosis, justifying universal prophylaxis with subcutaneous heparin or pneumatic compression boots.22 Gastric mucosal protection also should be provided for ventilated patients.23 Proton pump inhibitors, H2 receptor blockers, carafate, and antacids have been used to protect the stomach. Whether continuous feeding of the gut, which usually neutralizes pH, obviates the need for prophylaxis remains unclear. Ventilated patients often require sedatives and narcotics to maintain their comfort and synchrony with the ventilator. However, these medications have a cost. If overused, they delay patients' return to consciousness and full respiratory drive, thereby prolonging the duration of mechanical ventilation. One method of reducing the amount of sedative is to titrate ventilator volumes/flows/rates/modes to patient comfort before resorting to large doses of sedatives and narcotics. When these medications are used, they should be used on an “as needed” basis,24 or when a continuous infusion is used, a period of daily awakening should be encouraged.25

Item 3: Recognize Readiness—the Logistics of Liberation

Weaning implies gradual, rather than rapid, withdrawal of ventilatory assistance. This word suggests that the mechanism for separating the patient and the ventilator successfully resides in the machine next to the patient rather than in the patient himself or herself. The remainder of this chapter will emphasize a very simple truism that “many patients do not require repetitive measurements or involved plans for discontinuation of mechanical ventilation—if intercurrent problems causing respiratory muscle weakness or abnormal lung mechanics have not arisen, the most easily effected plan for discontinuation of mechanical ventilation is to wheel the machine out of the room.”26 Thus liberation more accurately describes the process by which most patients are freed from the ventilator. In patients who cannot breathe independently, liberation will be possible only after the patient is treated and recovers.

Historically, weaning parameters have been used to predict patients' ability to breathe without the ventilator. However, despite decades of research, no weaning parameter has predictive accuracy sufficient to be used exclusively to make liberation decisions.27,28 Moreover, it seems wasted effort to spend time trying to predict whether a patient can breathe without the ventilator when the question can be answered directly with a trial of spontaneous breathing. Accordingly, we do not use weaning parameters routinely to make liberation decisions. In the last edition of this book we dedicated a great deal of space to discussing weaning parameters. Now we provide instead references to demonstrate why clinicians simply should perform spontaneous breathing trials (SBTs) in most of their patients to determine their candidacy for extubation.29–46

To consider a patient for an SBT, he or she should be stable hemodynamically, without evidence of myocardial ischemia and, preferably, not excessively tachycardic. Oxygenation must be adequate,47 and the PaO2/FiO2 should be greater than 120 (i.e., PaO2 >60 mm Hg on a concentration of oxygen less than or equal to 50%) at nominal levels of PEEP (PEEP ≤5.0 cm H2O). If these two criteria are met, then attention turns to determining whether the patient can ventilate and oxygenate without the machine (Fig. 44-3).

Preparation

Sedatives and narcotics should be discontinued several hours before a breathing trial to reduce the likelihood of inadequate drive to breathe. Indeed, while several studies have indicated that protocol-driven approaches to weaning have identified patients earlier for extubation than routine care by clinicians on a day-to-day basis, other investigations protocolizing a daily interruption of sedatives have been associated with a decreased time of mechanical ventilation and diminished time in the ICU.25 It is likely that these are two tightly linked ICU activities—sedation and mechanical ventilatory support—and that clinicians are not extremely adept at identifying prospectively when patients do not need one or the other. A simple approach to this problem is to discontinue sedatives frequently and, in those patients awakening in a stable condition, to perform an SBT.

Ventilator Mode and Duration of the SBT

Pressure support, continuous positive airway pressure (CPAP), and T-piece trials are the most common methods used to test readiness for liberation from mechanical ventilation. The choice of whether to use pressure support or to allow the patient to breathe without inspiratory assistance through the endotracheal tube remains a point of physician style. Some data suggest that the endotracheal tube can increase resistive load significantly,48 which theoretically could contribute to failure. Consequently, many investigators have suggested that a pressure support of 5 to 8 cm H2O can be used to counter this tube-related load. Recent studies have suggested, however, that pressure support intended to overcome tube resistance unloads the respiratory muscles compared with extubated breathing. These data suggest that T-piece breathing most closely approximates the work of unassisted breathing after extubation.49,50 Nonetheless, one study suggested that the rate of extubation failure following successful trials of pressure support equal to 8 cm H2O was not greater than following successful T-piece trials.51 Another study found similar rates of successful extubation following trials of pressure support of 7 cm H2O versus T piece.52

The first SBT need be only 30 minutes53 because extending the trial longer does not enhance the clinician's ability to assess the patient's readiness for extubation. The proper duration of subsequent SBTs in those who fail has not been studied (30 to 120 minutes generally are used). Thus the available data suggest that patients should be considered for a trial of extubation after a successful trial (30 to 120 minutes) of either T piece, CPAP, or pressure support of 5 to 8 cm H2O.

The Patient Who Fails Initial SBTs

Failure of an SBT is most often a clinical diagnosis. The clinical signs of failure include rapid-shallow breathing, tachycardia (>110 beats per minute), hypertension (increment of more than 20 mm Hg), mental status changes, and subjective distress. These signs result from (1) decrements in gas exchange, (2) cardiovascular events, and/or (3) other (noncardiopulmonary) issues. When a patient fails an SBT, clinicians frequently focus on specific ventilator regimes aimed at improving respiratory muscle function. Although some studies in animal models have suggested that respiratory muscle exercise may be helpful,54 to date no study has suggested that the ventilator can be used to expedite the recovery from respiratory failure. If “exercise” is attempted, all attempts should be made to avoid exhausting the patient (generally accomplished if respiratory rates stay below 30/min). However, instead of concentrating on the ventilator, the clinician should turn attention to treatable factors underlying the patient's respiratory failure (Table 44-3). Some traditional weaning parameters are helpful in making this determination. For example, a markedly reduced maximum negative inspired pressure suggests respiratory muscle weakness. An elevated rapid-shallow breathing index (the respiratory rate per minute divided by the tidal volume in liters) suggests imbalance of respiratory muscle load and capacity. Reasons for abnormalities then can be examined by systematic diagnostic and therapeutic interventions.

Analyzing Failure of an SBT

Acute hypercapnia during weaning frequently results from an imbalance between respiratory pump capacity and load. Normal individuals who are subjected to resistance loading exhibit rapid-shallow breathing as a sign of impending respiratory failure.55,56 When the patient is clearly failing an SBT, it should be discontinued (even before a blood gas analysis demonstrates gas exchange abnormalities) because patients who are “weaned” to exhaustion frequently take several days to recover adequately to resume trials.

Many factors may reduce respiratory muscle strength or increase respiratory muscle loads in critically ill patients. In patients who fail due to strength-load imbalance, we quantify neuromuscular function and the three elements of respiratory muscle load (resistance, elastance, and minute volume; see above) each day so as to define/treat reversible elements. Critical illness frequently is associated with a catabolic state, malnutrition, and electrolyte deficiencies that can contribute to respiratory muscle weakness. Some patients recovering from severe illness develop a polyneuropathy57–60 that has been hypothesized to contribute to reduced respiratory pump function and prolonged need for mechanical ventilation.58,59 Sepsis occurs frequently in critically ill patients and is a relatively common reason for weaning failure. Numerous studies suggest that sepsis has a negative impact on respiratory muscle function,61–65 and this effect is mediated by cytokines.63,64 In addition, treatments that are used commonly in critically ill patients, such as administration of corticosteroids66–68 and neuromuscular blockers,69–71 also negatively affect respiratory muscle function. Mechanical ventilation itself may contribute to disuse atrophy of the respiratory muscles, thus contributing to the pathogenesis of prolonged ventilator dependence.54,72 Bronchospasm and increased airway secretions frequently contribute to resistive loading of the respiratory muscles. Elevated airway resistance greater than 15 cm H2O/L per second frequently can be reversed by removing excessive airway secretions73 or treating with aerosolized bronchodilators.74 If resistance remains greater than 15 cm H2O/L per second despite bronchodilators in a patient who repeatedly fails to wean, a therapeutic trial of corticosteroids may be helpful.

Finally, numerous factors contribute to increased respiratory system elastance, ranging from acute lung injury to abdominal distention. Pulmonary edema and pneumonia are common reversible causes of increased lung elastance. Occult PEEP also increases elastic load75 and can contribute to respiratory muscle fatigue by increasing the work of assisted76 and unassisted breathing. Increased minute volumes associated with lung injury, hypermetabolic states of critical illness (e.g., sepsis), pulmonary embolism, and overfeeding could contribute to dynamic hyperinflation during the recovery process, thus increasing mechanical loads on the recovering respiratory muscles.

Acute hypercapnia does not necessarily connote weaning failure. In three relatively common situations, hypercapnia during an SBT trial is not accompanied by physiologic decompensation and does not signal ventilatory failure. First, occasional patients with chronic hypercapnia are hyperventilated iatrogenically to a normal PCO2, and over the course of days in the ICU, the bicarbonate concentration also decreases to normal. When these patients resume spontaneous breathing, their PCO2 rises to their baseline level, leading to an acute respiratory acidosis. The clinician must decipher whether the acute acidosis is arising from strength-load imbalance, in which case the patient usually shows other signs of failure, or from preceding iatrogenic hyperventilation. Similarly, when some patients with chronic hypercapnic respiratory failure are given a high fraction of inspired oxygen, they acutely retain CO2 in the absence of strength-load imbalance.77–79 Reduction in the supplemental oxygen to yield a saturation of 90% to 92% may reverse this iatrogenic cause of hypercapnia. Finally, some patients who have primary metabolic alkalosis will have compensatory hypoventilation and hypercapnia.

Hypoxemia (PaO2 <60 mm Hg or saturation less than 90% on 50% inspired oxygen) can occur during weaning for several reasons. Old age, obesity, and recumbency predispose to a lower functional residual capacity (FRC), which can contribute to atelectasis and subsequent hypoxemia during the transition to unassisted breathing. Thus patients should be sitting at greater than a 30-degree angle during spontaneous breathing. Respiratory muscle weakness and sedatives or narcotics may lead to shallow breathing and atelectasis. Lung injury, a common complication of critical illness, is associated with surfactant depletion and an increased propensity for atelectasis during the withdrawal of positive-pressure ventilation. Thus many pulmonary factors can contribute to hypoxemia during SBTs. Hypoxemia also can result from cardiovascular changes during weaning.

The transition to unassisted breathing is associated with centralization of blood and increased left heart afterload, which may predispose patients with baseline left ventricular dysfunction to develop cardiogenic pulmonary edema.6–15 SBTs also may increase the level of circulating catechols,80 which could induce increases in heart rate, blood pressure, and arrhythmias. Tachycardia during weaning frequently is a sign of inadequate cardiopulmonary reserve and subsequent weaning failure. Moreover, in patients with coronary artery disease, if mechanical loads and catechol-induced changes are sufficient, weaning can trigger cardiac ischemia that could contribute to the clinical signs of weaning failure81–84 (see Fig. 44-2). In a study of 93 medical patients being weaned from mechanical ventilation, ST-segment changes were noted in 6% of all patients and in 10% of those with a preceding history of coronary artery disease. Moreover, weaning-related ischemia tended to increase the risk of weaning failure.84

In patients who are clinically hypervolemic or who have left ventricular dysfunction, pre-emptive diuresis and nitrate therapy can be helpful in attenuating acute increases in preload during SBTs. Continuous monitoring of ST segments and treatment with additional nitrates also may be helpful in patients who experience ischemia during weaning.84

Other Reasons for SBT Failure

Even though most weaning failures are of cardiopulmonary etiology, other exogenous factors also may contribute to failure.

Ventilator and Circuit Elements

The ventilator and its circuitry can contribute to weaning failure by two mechanisms: (1) by increasing respiratory loads enough during an SBT to fatigue the respiratory muscles and (2) by imposing significant respiratory muscle work during “rest” periods. The resistance of the endotracheal tube increases with time, and this increase occasionally can be of sufficient magnitude to impede weaning. The ventilator circuit provides increased dead space and in some modes (for older ventilators) can require excessive work to trigger a “sticky” demand valve. Some studies suggest that flow triggering may reduce the ventilator-imposed work of breathing compared with pressure-triggered modalities.85,86 Thus the ventilator circuit can load87,88 and covertly fatigue the respiratory muscles when patients are presumed to be “resting,” as well as when they are weaning. Patient-ventilator synchrony during “rest” periods reduces the likelihood that the ventilator is contributing to weaning failure.87,89,90 Irregular pressure-volume curves or frequent, large esophageal or intravascular pressure fluctuations during assisted ventilation may help to identify patients who are working hard on the ventilator.91 Empirical manipulation of inspiratory flow rates, waveforms, and triggering mechanisms may aid in improving patient-ventilator synchrony.86 Finally, in patients with obstructive lung disease, intrinsic PEEP may increase significantly the work required to trigger ventilator-supported breaths (so-called trigger asynchrony). In selected patients, addition of applied PEEP to nearly match intrinsic PEEP can reduce this ventilator-induced load.76

Psychological and Housekeeping Issues

Subjective distress, including anxious appearance and sweating, are nonspecific signs of weaning failure. Anxiety during breathing trials may manifest as rapid-shallow breathing, tachycardia, and hypertension that are interpreted as weaning failure by caregivers. Bedside personnel should explain the weaning process; so-called verbal anesthesia is effective in some patients. The use of sedatives to treat anxiety must be undertaken with caution because most of these drugs also depress respiratory function. Haloperidol or very low doses of benzodiazepines can be used to facilitate comfort in these exceptional patients. However, subjective distress most frequently signals that the cardiopulmonary system requires additional repair before resuming unassisted breathing.

The effect of re-establishing patient day-night cycles remains to be well studied.92–94 In patients who fail initial trials of breathing, we are careful to ensure sleep at night and wakefulness, if not exercise, during the day.

Liberation Strategies

Many intensivists have reasoned that by reducing ventilatory support gradually, the respiratory muscles exercise at subfatiguing loads, leading to gradual improvement in function. Some studies have suggested that respiratory exercises (repetitions of low-load resistive breathing) can lead to successful extubation in patients who have failed previously.95 However, there are as yet no studies proving that respiratory muscle training, through the use of graded withdrawal of ventilatory support, hastens the recovery to unassisted breathing.

Two recent studies have assessed the role of “weaning” strategies in expediting liberation of the subset of those who fail initial SBTs. Brochard and colleagues51 studied 456 medical-surgical patients being considered for weaning, of whom 347 (76%) were extubated successfully on the first day of weaning. One hundred and nine patients who failed an initial SBT were randomized to be weaned by one of three strategies: (1) T-piece trials of increasing length until 2 hours could be tolerated, (2) intermittent mandatory ventilation with attempted reductions of 2 to 4 breaths per minute twice a day until 4 breaths per minute could be tolerated, and (3) pressure-support ventilation (PSV) with attempted reductions of 2 to 4 cm H2O twice a day until 8 cm H2O could be tolerated. Patients randomized to the three strategies were similar with regard to disease severity and duration of ventilation before weaning. Patients assigned to T-piece were placed on assist control overnight, whereas synchronized intermittent mandatory ventilation (SIMV) patients remained on an unspecified SIMV rate, and PSV patients remained at an unspecified level of pressure support. There was no difference in the duration of weaning between the T-piece and SIMV groups, but PSV led to significantly shorter weaning compared with the combined T-piece and SIMV cohorts. Interestingly, PSV and T-piece were not compared directly. These authors concluded that “the outcome of weaning from mechanical ventilation was influenced by the ventilatory strategy chosen, and the use of PSV resulted in significant improvement compared with other strictly defined weaning protocols using T-piece or SIMV.”

Esteban and colleagues96 performed a similar study of 546 medical-surgical patients, 416 (76%) of whom were extubated successfully on their first day of weaning. The 130 patients who failed were randomized to undergo weaning by (1) once-a-day T-piece trial, (2) two or more T-piece or CPAP trials each day as tolerated, (3) PSV with attempts at reduction of 2 to 4 cm H2O at least twice a day, and (4) SIMV with attempts at reducing 2 to 4 breaths per minute at least twice a day. Patients assigned to the four groups were similar with regard to demographic characteristics, acuity of illness, and a number of cardiopulmonary variables. Duration of ventilation before weaning was shorter in the SIMV group than in other groups. The mode of ventilation at night was not specified. The weaning success rate was significantly better with once-daily T-piece trials than for PSV and SIMV. Twice-daily T-piece trials were not significantly better, and PSV was not superior to SIMV. The median duration of weaning was 5 days for SIMV, 4 days for PSV, and 3 days for the T-piece regimens. These authors concluded that “a once-daily trial of spontaneous breathing led to extubation about three times more quickly than intermittent mandatory ventilation and about twice as quickly as pressure-support ventilation.”

The statistical methods used in these studies warrant some comment. The first study51 asserts the superiority of pressure-support weaning to T-piece but does not compare these directly. The second study96 presents “adjusted” rates of successful weaning—the nature of the adjustments and their impact on the results are unclear because raw data were not presented. However, several important conclusions can be drawn from these relatively large studies. First, and most important, most patients can be extubated successfully on the first day that physicians recognize readiness after a brief trial (30 to 120 minutes) of breathing through a T-piece. Weaning is not necessary for most patients. Second, both studies suggest that in patients who have failed an initial T-piece trial, SIMV weaning prolongs the duration of mechanical ventilation. These conclusions depend on the precise algorithms used in the studies—it is possible that if a more “aggressive” SIMV algorithm were used, no difference would be noted. However, if one considers the results of these studies together, choices of ventilator mode do not appear to have a major impact in speeding the recovery to unassisted breathing. Clinicians frequently use SIMV or PSV as the initial approach before attempting a T-piece trial. Since most patients are extubated successfully after a brief T-piece trial, such an approach is likely to prolong mechanical ventilation for many patients. PSV weaning may be equivalent to T-piece weaning in carefully controlled protocols. However, in our experience, in busy ICUs, respiratory therapists and nurses frequently are unable to “keep up” with the demands of such protocols—so pressure-support withdrawal protocols have the risk of prolonging ventilation unless followed strictly.

There is an important distinction between liberation from mechanical ventilation and extubation. Once a patient has been liberated, i.e., has passed the SBT, he or she does not need the ventilator. The endotracheal tube, aside from providing the avenue for PPV, allows removal of secretions for patients whose airway protective mechanisms have been altered by disease. Accordingly, after a patient has performed a successful trial of unassisted breathing (is liberated from the ventilator), the physician must determine whether the patient still needs the artificial airway.

One consideration is whether the patient has an upper airway lesion that would collapse to a critically small size after extubation. Patients at risk of this problem include those who were initially intubated for upper airway stenosis and stridor and those who have had a traumatic or prolonged intubation.97 A number of studies have suggested that the ability to breathe around a deflated endotracheal tube cuff98–100 or a cuff leak of greater than 110 mL during volume-cycled ventilation100 reduces the likelihood of postextubation failure. However, lack of a cuff leak does not absolutely predict extubation failure.

The patient's mental status, airway protective mechanisms, ability to cough, and amount of secretions are also determinants of extubation outcome. Patients with cough peak flows of 60 L/min or less are five times as likely to fail a trial of extubation (TOE) compared with those with cough peak flows of more than 60 L/min. Those who cannot cough and wet a white card placed 1 to 2 cm from the end of the open endotracheal tube are three times as likely to fail a TOE. Those who require endotracheal suctioning more than every 2 hours are also at increased risk of extubation failure.47,101 Thus, in general, we are disinclined to extubate patients if they have excessive or tenacious secretions and a weak cough or are not cooperative enough to aid in their own pulmonary toilet (deep breathing and expectoration of secretions). Patients with severe cerebral vascular accidents (CVAs) frequently present with this constellation. It remains unclear as to whether early elective tracheostomy to prevent aspiration and aid in pulmonary toilet is superior to a trial of extubation. When the cause of impaired airway protection is thought to be reversible, treating and attempting a TOE at a later time are reasonable. Recent data suggest that swallowing is abnormal in more than 30% of patients following extubation.102 Accordingly, all patients, especially those with altered mentation or CVA, should be observed carefully after extubation, and formal swallowing assessment is advisable for many patients.

Postextubation stridor arising from upper airway edema is fairly common after extubation. Pulmonary edema can develop in some of these patients103 because large negative intrathoracic pressures during inspiration can increase left ventricular afterload dramatically.11 Nebulized racemic epinephrine and parenteral corticosteroids have been used for the treatment of airway edema but have not been studied systematically. Heliox or mask CPAP may be used to reduce upper airway resistance temporarily in selected patients who do not require immediate reintubation.104

Noninvasive face-mask positive-pressure ventilation (NIPPV) may prevent the need for reintubation in selected patients who appear to be failing immediately after extubation.105–107 One study prospectively randomized 39 patients with COPD and normal sensorium who had failed a T-piece trial to pressure-support weaning or extubation to NIPPV. Patients assigned to NIPPV were more likely to be liberated from mechanical ventilation and succeeded 5 days sooner than those who were weaned using pressure support.107 In a multicenter trial of 221 patients failing extubation (only 10% having COPD), NIV was ineffective in preventing reintubation and did not reduce mortality.107a Thus NIPPV remains unproved for extubation failure, but may be useful in cooperative patients who appear to be failing extubation and who do not require immediate reintubation for impending respiratory arrest.

Deep-breathing exercises and incentive spirometry after extubation may help to prevent atelectasis in at-risk patients. Intermittent positive-pressure breathing treatments may be helpful to reduce atelectasis in patients who have not recovered adequately to cough and deep breathe. Continuous ST-segment monitoring can identify some patients with myocardial ischemia; early recognition and treatment may reduce the likelihood of recurrent respiratory failure. There are few data to support these suggestions, but common sense may not require prospective validation.

Most experienced clinicians have treated patients whose SBTs suggested that they would fail but who were liberated and extubated successfully nonetheless. When the numbers look bad but the patient looks good or there is concern that the presence of the endotracheal tube is responsible for weaning failure, it is reasonable to perform a careful TOE. The following are among the clinical situations that could prompt consideration for a TOE in such patients:

1. When an endotracheal tube has been in place for more than 7 days. Endotracheal tube resistance increases with time and can contribute to failed SBTs.

2. When the patient experiences repeated episodes of bronchospasm on awakening from sedation. The endotracheal tube can cause reflex bronchospasm in some individuals.

3. When patients become overwhelmingly anxious when awakened to breathe through the endotracheal tube and the amount of sedative required to make them comfortable causes hypoventilation. We are particularly careful to ensure that cardiopulmonary reasons for failure have been reversed in these patients.

4. When patients with restrictive chest wall disease (e.g., obesity) repeatedly desaturate every time PEEP is decreased to less than 10 cm H2O. Some obese patients require more than 5 cm H2O to prevent atelectasis while intubated yet maintain adequate oxygenation when extubated.

5. When patients with severe restrictive or obstructive lung disease breathe rapidly and shallowly (a rapid-shallow breathing index or frequency/Vt >125 breaths per minute per liter). For some end-stage patients, rapid-shallow breathing is their chronic baseline.

In these relatively rare situations, extubation should not be performed casually. We consider breaking the rules outlined in this chapter only after numerous failed trials of unassisted breathing and after treating reversible causes of failure. The clinical risks associated with failure and reintubation must be weighed against those of continued mechanical ventilation. We extubate these unusual patients with personnel who are skilled at endotracheal intubation nearby, should reintubation be required. In addition, we ensure ready access to NIPPV, which may avert the need for reintubation in carefully selected patients.

At many institutions, patients remain bound to ventilators for longer than necessary because clinicians either do not obtain weaning parameters in a timely manner or are hesitant to extubate even when parameters and breathing trials are favorable. Mechanical ventilation can cause complications.15–17 Conversely, there is a small risk associated with extubation failure. The relative risks of each vary from patient to patient and from institution to institution. A simple approach can minimize both risks. Each day the clinician should ask: Can my patient be liberated from the ventilator? The use of interdisciplinary weaning teams108–110 or respiratory therapist–driven protocols may expedite successful liberation by actively addressing this question each day.

Ely and colleagues studied 300 mechanically ventilated adult medical patients, of whom 149 were prospectively randomized to receive algorithm-guided attention to weaning, and the remaining 151 received “routine” care.110 All patients were assessed daily to delineate recovery to adequate oxygenation (PaO2/FiO2 >200 on PEEP ≤5 cm H2O), hemodynamic stability, and a frequency/Vt of 105 breaths per minute per liter or less (measured on CPAP). When all these criteria were met, a note was left in the charts of patients assigned to the treatment group suggesting that they were ready for liberation. Overall, patients whose care was guided by these criteria were liberated 1.5 days more rapidly than patients in the control group. The median weaning time after satisfaction of the criteria was 2 days shorter in the treatment group. This reduction in ventilator days translated to a trend toward reduced ICU length of stay and a significant reduction (by $5000) in ICU costs of care. Interestingly, median weaning time from satisfaction of the screening criteria was 1 day in the treatment group, suggesting that further reductions could have been realized if the gentle “reminder” led to a prompt trial of unassisted breathing. Other studies47 have suggested that the oxygenation criteria used in this study may be conservative (e.g., PaO2 ≥80 mm Hg on 40% oxygen). In addition, roughly 50% of patients with a frequency/Vt of 100 to 125 breaths per minute per liter can be extubated successfully.32 Thus additional reductions in duration of mechanical ventilation can potentially be achieved using a more aggressive algorithm (Fig. 44-3). Note also that these results cannot necessarily be generalized to all institutions because “routine care” (the control group in this study) likely varies considerably among hospitals. However, this study strongly suggests that many patients are inappropriately retained on mechanical ventilators. Other studies have applied very different algorithms to achieve significant reductions in duration of ventilation.18,111 However, they all have one thing in common: They substitute a program of daily systematic scrutiny of readiness for breathing for the individual variation occurring in unstructured care systems. Whether achieved by protocol or by individual clinician perseverance, we believe that patients can be liberated from mechanical ventilation more expeditiously if they are screened on a daily basis for this potential event.

Our approach can be summarized as follows: If the patient is hemodynamically stable, awake, and triggering the ventilator with a PaO2/FiO2 greater than 120 on PEEP of 5.0 cm H2O or less, we perform an SBT. Alternatively, patients can be screened with a 1- 2-minute T-piece trial using the frequency/Vt to guide whether to lengthen the trial (as in the Ely and colleagues study).110 We prefer pressure support at a low level or T piece on 50% oxygen for 30 minutes (60 minutes after a first-day failure). If the patient remains comfortable, with an oxygen saturation greater than 90% and without tachycardia or hypertension, and the breathing rate remains less than 35/min at 30 minutes, we will obtain an arterial blood gas and consider a TOE using independent judgments to gauge the patient's ability to do without the endotracheal tube itself (see above). For patients who fail an initial SBT, we aggressively assess the pathogenesis of their failure and perform daily trials of either T piece or pressure support of less than 7 cm H2O until the patients can tolerate a greater than 30-minute trial of unassisted breathing and then proceed to liberation from the ventilator and assessment for extubation.