Chapter 5. Weaning and Extubation

Mechanical ventilation consists of two parts: the ventilator and the delivery device, most commonly an endotracheal tube (ETT). Similarly, there are two parts to liberation from mechanical ventilation: reducing and then eliminating the assistance provided by the ventilator and removing the ETT.

Although much attention is paid to when to intubate patients, significantly less attention is paid to when and how to extubate them. For those accustomed to literature searches using databases such as Ovid, there is not even a MeSH heading for “extubate” or “extubation.” There is similarly less guidance on how to reduce the patient's reliance on the ventilator, that is, weaning. While weaning is—or should be—a precursor to extubation, the reverse is not always true, that is, a patient not ready for extubation still may be weaned. For example, a patient with a penetrating neck injury with a rapidly expanding hematoma requires an ETT for airway protection, but may only require minimal, if any, support from the ventilator. Either way, given the risk and expense of intubation and mechanical ventilation,1,2 there should be no delay in ventilator weaning and patients should be extubated as soon as possible.

Before deciding on whether to wean or extubate a particular patient, one must recall why the patient was intubated in the first place. Many patients are intubated for respiratory failure, either for a primary pulmonary problem (e.g., pneumonia), a problem with another organ system (e.g., myocardial infarction, fluid overload from renal failure), or a systemic problem (e.g., sepsis). Other patients are intubated for airway protection, either from altered mental status (e.g., drug overdose) or because of impending airway collapse (e.g., anaphylaxis, smoke inhalation, trauma). The reason for intubation will have a direct bearing on when, and if, a patient may be weaned or extubated.

The majority of patients extubated in the ED are those who either self-extubate or were intubated for a state reversed with time in the ED, for example, drug overdose. A recent study showed that ED extubation in such conditions can be safe and can successfully reduce the need for additional resources such as an ICU bed, or even admission altogether.3 Otherwise, it falls to the intensivist to wean and extubate the patient.

Clinical progress and, in fact, early research in the area support the progression from weaning to extubation. Although recent efforts have questioned the approach albeit with equivocal results,4–6 traditionally extubation is the last step in liberating a patient from mechanical ventilation.

Since a failed extubation (defined as an extubation followed by an unplanned reintubation) carries significant mortality and morbidity,7,8 an assessment should be taken to determine which patients will remain extubated and which will fail and require emergent reintubation. For the latter, instead of extubation, further treatment of the patient's underlying condition(s) or additional weaning steps should be undertaken, with continual reassessment for readiness to extubate. The former group should be extubated as soon as possible. So how, then, can a patient's readiness for extubation be ascertained?

The first step is to recall why the patient was intubated. For patients intubated to protect their airway or facilitate evaluation such as physical exam and imaging studies during an altered mental state, it seems reasonable to wait until the patients have had return of their baseline mental status. These patients, generally intubated for only short periods of time, may not need any assessment of their readiness to extubate other than an assessment to ensure return of their baseline mental status. These are the patients most commonly electively extubated in the ED.3

Except for those few patients extubated in the ED, a significant amount of time may elapse before a patient's clinical condition improves to the point where he or she may tolerate weaning and extubation. A variety of different parameters have been studied to determine which, if any, predicted a successful extubation.9 Depending on the reason for intubation, one test with one parameter has generally been shown superior to other tests. For those patients intubated for respiratory failure, the spontaneous breathing trial (SBT) with calculation of the rapid spontaneous breathing index (RSBI) has been shown to most closely correlate with successful liberation from mechanical ventilation. For those patients intubated for airway protection, the cuff-leak test has similarly been shown to be predictive. Furthermore, many practitioners routinely perform both tests: an SBT to assess the patient's ability to breathe without artificial support and then a cuff-leak test to assess the patient's natural airway.

An SBT is, just as its name implies, a clinically applied test to see if and how a patient will breathe on his or her own. Artificial ventilatory support (but not the ETT) is removed, and the pattern of the patient's ventilation is observed. A “successful” test is one that predicts successful extubation, where “successful extubation” means a patient does not require reintubation for the same condition within 24 hours.

Over time, many parameters have been investigated to see which best defines a successful SBT. In other words, which parameter during an SBT best predicts successful extubation. The RSBI (often pronounced “risby”), also referred to as the Yang–Tobin Index or, simply, the Tobin Index, has been consistently proven to be the best predictor of which patients may be successfully extubated. The RSBI is the respiratory frequency divided by the tidal volume (TV) while a patient is breathing spontaneously, as follows:

where f is the number of breaths per minute, VT the average TV per breath in liters, and the RSBI expressed in breaths per minute per liter. It can be seen from the equation that rapid, shallow breaths (“panting”) such as occur with respiratory distress would result in a very high RSBI. Conversely, fewer, deeper breaths result in a lower RSBI. In a neurologically intact patient, therefore, lower scores are better.

A RSBI of ≤105 predicts successful extubation with a sensitivity of 97%, specificity of 65%, positive predictive value of 78%, and a negative predictive value of 95%. That means patients ready for extubation have a 97% chance of having a RSBI ≤105 (in other words, almost every patient ready to be extubated will have a RSBI ≤105). Unfortunately, the specificity of 65% means that 35% of patients who have RSBI <105 will still fail extubation. And therein lies the rub. Even the RSBI—the best proven predictor to date—will lead clinicians to extubate many patients who will fail, and to leave intubated a few patients who could otherwise be liberated from mechanical intubation.

Subsequent investigations into the RSBI have yielded varying thresholds, but in general a value ≤105 correlates with an 80% chance of successful extubation (extubation not requiring reintubation for the same cause within 24 hours) and a value >105 correlates with a 95% chance of the patient failing extubation.10 Some practitioners prefer to use a RSBI value of 100 considering it an easier number to remember.

It is important to keep in mind a RSBI is only valid during an SBT. Historically, SBTs were performed by disconnecting the patient from the ventilator and placing them on a respiratory circuit, often referred to as a T-piece due to the circuit's shape (see Figures 5-1 and 5-2). This setup, however, is rarely, if ever, still encountered in clinical practice. There are two reasons for this, one physiologic and one practical.

Physiologically, the T-piece circuit increases the amount of dead-space ventilation (see Figure 5-1). Unlike in vivo dead space, the “bleeding” of oxygen into the T-piece circuit ensures that the circuit's dead space does not adversely affect fraction of inspired oxygen (FiO2) actually reaching the alveoli. That is to say, the oxygen bled into the T-piece circuit clears out expired gases and provides a reservoir for inhalations faster than oxygen can be delivered. However, exactly like in vivo dead space, the T-piece circuit adds resistance to gas flow that increases the work of breathing.

In one respect, this can be seen as an advantage. A patient capable of breathing through a T-piece circuit—which requires more effort than if not intubated at all—should successfully tolerate extubation. Conversely, this means there are patients who cannot tolerate breathing through a T-piece but who would tolerate extubation. Further proving this point, during investigation into shorter-duration SBTs, Esteban et al showed that patients subjected to longer durations breathing through a T-piece circuit actually had longer ICU and hospital LOS.11 In other words, the “bar” of this test is set too high, so high, in fact, that the test itself is harmful.

The practical disadvantage of T-piece trials involves the T-piece circuit itself. First, it requires extra equipment. Second, it requires disconnection from the ventilator. Given the built-in monitoring and safety features of modern ventilators (e.g., apnea monitoring), disconnection from the ventilator in favor of a circuit with no intrinsic monitoring introduces risk and a chance for medical error.

Therefore, in clinical practice, T-piece breathing trials have been eliminated in favor of SBTs with the patient still connected to the ventilator. This allows the same RSBI to be calculated while maintaining the monitoring and safety features of the modern-day ventilator. Moreover, most modern ventilators will automatically calculate the patient's RSBI, although care must be taken to ensure the RSBI is calculated over the desired time period, and is not “just a number” that the ventilator calculates for some preceding, random time period (such as over only the preceding minute).

Note that a RSBI is only valid if the patient is breathing completely spontaneously. If the patient is still receiving machine-delivered breaths, the RSBI is not valid, even if the ventilator reports a score.

Given that SBTs are better conducted with the patient remaining connected to the ventilator, an improvement would be if the ventilator could also be used to overcome the physiologic disadvantage of increased dead space causing increased resistance and therefore work of breathing. By setting a low level of pressure support (PS), the ventilator can, indeed, assist in making the test better by helping to overcome the extra work of breathing imposed by the ventilator circuit.12,13

Some ventilators, in fact, have a feature to do this automatically. For ventilators so equipped, “automatic tube compensation” (ATC) allows the ventilator to automatically and dynamically adjust its own support of a patient's breathing to provide the exact amount of support necessary to overcome the circuit. This may, in fact, be more accurate than simply setting a static level of support since resistance of the ventilator circuit is a function not only of length and cross-section (which are relatively constant) but also of velocity (flow) and acceleration (change in flow), both of which vary across the respiratory cycle itself, from breath to breath, and as the patient's clinical condition improves.

The remaining question is for how long to perform an SBT. Traditionally SBTs were performed for 120 minutes.2,14,15 Subsequent investigation showed that for patients who have not yet been given a weaning or extubation trial, if a patient was to fail a 120-minute SBT, the failure was going to occur within 30 minutes; contrapositively, successful completion of a 30-minute SBT predicted successful completion of a 120-minute SBT and therefore successful extubation.11

Furthermore, 15% of patients of Esteban et al who required reintubation did so because of upper airway obstruction, for which there is now a good test, the cuff-leak test, which is detailed later in this chapter. Had patients of Esteban et al with upper airway obstruction been identified and not extubated, SBTs would have been shown to be even more successful.11

Weaning is the act of reducing ventilatory support in preparation for liberation from mechanical ventilation. Now that specific, objective criteria (RSBI ≤105) have been described, attention must be given to the methods to bring the patient to that state, specifically, to the ventilatory and adjunct strategies that can be employed to improve a patient's respiratory status to one with a RSBI ≤105, when the patient does not require any additional ventilatory assistance.

First consider, independently, the FiO2, and when, after intubation, it may be reduced. Until the mid-1990s, the variables commonly available to evaluate artificial ventilation were those obtained by an arterial blood gas (ABG): pH, PaO2, and Paco2. When considering how much oxygen was being delivered to the systemic arteries and whether or not FiO2 can be reduced, the salient value was the PaO2. But with the widespread use of pulse oximetry, it is now possible to know how much oxygen is being delivered to the systemic arteries by the continuous, noninvasive, bedside measurement of the oxygen saturation of arterial hemoglobin (SaO2). Recall the equation of oxygen content of arterial blood16:

CaO2 = [1.34 × Hgb × SaO2] + [0.003 × PaO2]

where 1.34 is the oxygen binding capacity of hemoglobin (mL O2 /g Hgb), hgb the concentration of hemoglobin in blood (g/dL), SaO2 the percent hemoglobin saturated with oxygen, 0.003 the solubility coefficient of oxygen in water at PO2 of 1 mm Hg (mL O2/100 mL water/mm Hg), PaO2 the partial pressure of oxygen (mm Hg), and CaO2 the oxygen content of arterial blood (g O2/dL plasma).

From the above equation, it can be seen that, except in cases of hyperbaric oxygen therapy, the partial pressure of oxygen is only a minimal contributor to the oxygen content of arterial blood and the term, therefore, may be removed from the equation. The equation then simplifies to:

CaO2 = 1.34 × hemoglobin × SaO2

From the above equation, it can be seen that for a given hemoglobin concentration, the amount of oxygen in arterial blood is directly and completely measured by pulse oximetry. It also shows that SaO2 (as determined by a pulse oximeter) is, in fact, a better measure of oxygen content than PaO2 (as reported on a blood gas result). For blood gas instrumentation that reports oxygen saturation results, it must be kept in mind that frequently the saturation number is calculated from the PaO2, and not measured directly as is done by a pulse oximeter.

Since SaO2 is a dynamic measure of the oxygen content of arterial blood, the ventilator's FiO2 can be titrated by monitoring pulse oximetry. In fact, given the dangers of hyperoxia17,18 and the ease, safety, and rapidity with which SaO2 is measured, FiO2 should be reduced as soon as possible after intubation, as tolerated, by monitoring the pulse oximeter.

The remaining parameters to wean will be the amount and frequency of ventilator support as well as how the ventilator delivers that support (i.e., the mode). The first two parameters determine minute ventilation, the adequacy of which can only be objectively ascertained by the ABG: specifically, the pH and Paco2. Still, unlike a continuous bedside pulse oximeter, ABGs are only snapshots in time. Fortunately, there is a monitor capable of continually measuring the patient's pH and, thereby, Paco2: the patient's own respiratory center.

It follows, then, as soon as a patient's condition sufficiently improves, sedation should be lightened enough to ensure comfort and cooperation with the ventilator (ventilatory synchrony) yet allow normal ventilatory drive. All that remains is to use a ventilatory mode that allows the patient to regulate his or her minute volume. This suggests that modes that allow the patient to both breathe spontaneously and regulate the volume of his or her spontaneous breaths will facilitate weaning.

Once a patient has achieved an adequate and comfortable respiratory status on the ventilator, parameters can then be reduced (e.g., fewer mandatory breaths per minute), to allow the patient to take over more and more of his or her own breathing. Finally, the patient can be moved to an entirely patient-controlled ventilatory mode, providing just enough support to overcome the resistance of the ETT and ventilator support.12,13 To achieve this goal is to achieve the maximal level of ventilatory weaning, proving the patient's pulmonary, cardiovascular, and other organ systems can tolerate extubation. Of course, this goal says nothing about extubation as a function of the patient's airway—itself reliant on airway anatomy and physiology (e.g., no swelling)—or the patient's ability to protect the airway, but it is the end point of weaning after which no further decrease in ventilatory support is either meaningful or appropriate.

Finding the optimal way to move a patient from initial ventilator settings to a state where the patient is ready for extubation is not trivial. Indeed, much work has been done on the topic.15

Weaning Methods

It would seem that the logical way to eventually remove support of any kind would be to decrease the amount of support and evaluate whether the subject tolerates the loss. With regards to ventilatory support, it is possible to decrease either, or both, the amount or frequency, of support, as well as additional parameters. Some approaches to weaning have involved simply removing the patient from the ventilator for a short period of time and, if tolerated, gradually increasing the length of time. One then establishes some value for time off the ventilator that, once reached, is assumed to indicate that the patient can function off the ventilator permanently. Often a cutoff of 2 hours has been used, but periods as long as 24 hours have been reported. This process, however, involves removing the patient from the ventilator, which, as discussed above, is not desirable. By using the ventilator for assistance, support can be gradually reduced or removed to facilitate a faster weaning.

The first consideration when using the ventilator to assist in weaning is the mode of ventilation. In 1994, Brochard et al were the first to examine the question of the optimal mode for weaning in a prospective randomized controlled trial (RCT).14 After excluding all patients who passed a single 2-hour T-piece SBT and were successfully extubated, they randomized patients to be weaned via gradually increasing SBTs on T-piece, synchronized intermittent mandatory ventilation (SIMV), or pressure support ventilation (PSV). They found that patients weaned fastest and most successfully using PSV. Furthermore, PSV yielded a shorter length of stay in the ICU and a trend toward lower all-cause mortality.

Conversely, in 1995, Esteban et al found that PSV actually increased weaning time,15 with SIMV being better, and once-daily SBTs on T-piece the best. However, unlike Brochard who only studied patient's who failed their first SBT or extubation attempt, Esteban studied all comers. This means Esteban's results are colored by the fact that 76% of his patient population was successfully extubated on the first attempt, suggesting a majority of his patients did not require weaning, per se, merely identification of those patients ready to be extubated on their first attempt. This crucial point was proven in a 1996 study by Ely et al that demonstrated that many patients who can be successfully extubated are not identified by the intensivist as such,2 and thus require objective evaluation and action based on positive performance on the evaluative measures. The only firm conclusion that can be drawn is that SBT is not the best weaning method, but rather is an indispensable test to identify patients as soon as they are ready for extubation.

Although denied in the discussion, Brochard's approach to SIMV potentially blunted the performance of SIMV or T-piece by using a protocol that limited how fast patients could be weaned on various modes. In other words, if SIMV did, in fact, lead to faster weaning in this study, it would not be apparent since the speed with which a patient could be weaned was limited by the study's protocol.

Finally, in both studies continuous positive airway pressure (CPAP) was allowed, but not required, which may confound the results of both studies. Since positive end-expiratory pressure (PEEP) of 5 cm H2O is generally considered physiologic, it would seem a possible methodological bias to have not included PEEP on all patients, or at least have it set and titrated based on protocol. Brochard does note that CPAP was set in most patients with chronic obstructive pulmonary disease (COPD) to facilitate patient triggering of the ventilator by ensuring pressure in the circuit equaled the patient's intrinsic PEEP, which decreases the inspiratory effort needed to drop the pressure sufficiently to trigger the ventilator.

For other less traditional modes, the mode itself may either not lend itself to weaning via eventual conversion to CPAP and PS or lend itself to a distinctive weaning methodology. As opposed to SIMV, which is usually programmed with volumes (e.g., volumes of 6–8 mL/kg), pressure-regulated modes function by lessening the pressures applied before lowering the mandatory rate. This can be done by lowering the mandatory-breath pressure and monitoring to ensure that adequate volumes are maintained. As pulmonary compliance, and therefore resultant volumes, increases, the pressure may be again reduced. At lower pressures, the mandatory rate can eventually be lowered.

In airway pressure release ventilation (APRV), it seems prudent to first lower Phigh until it is in a safe range below 20 mm Hg. At that point, Thigh may be increased, therefore increasing the time the patient remains at Phigh, which also decreases the number of releases, or mandatory breaths, per minute.19 This method is commonly known as the “drop and stretch” method, referring to dropping, or lowering, Phigh and stretching, or increasing, Thigh. In general, Phigh can be dropped in increments of 2 mm Hg, while Thigh can be increased in 0.5- to 2-second increments. Eventually Phigh will end up at levels of 5–12 mm Hg and, as Thigh is increased, CPAP is approximated. Of course, PS or ATC should also be applied, assisting the patient's spontaneous breaths in overcoming the resistance of the circuit, as in more traditional modes. Once these settings are achieved, the patient may be extubated directly from APRV or may be converted to CPAP/PS to calculate a RSBI.

Extubation is the act of removing the artificial ventilatory delivery device, which usually means removing the ETT. Traditionally ETTs are removed and the patients are left on room air, oxygen via nasal cannula or mask, or, more commonly, humidified oxygen via a simple mask. Recently, consideration has been given to extubating patients—earlier than once thought—and immediately applying noninvasive ventilation (NIV).

Extubation to Noninvasive Ventilation

When considering application of NIV, a distinction must be made between applying NIV after extubation immediately and applying NIV to a patient recently extubated but who is now having respiratory distress.

Planned Extubation to NIV

The mechanisms and advantages of NIV are discussed elsewhere in this text, and will not be repeated here. It should be noted that in most cases of home, ED, or other uses of NIV, the predominant modes do not include a mandatory rate. In other words, CPAP or CPAP/PS (which is also known as bilevel positive airway pressure [BiPAP]) is usually applied via face or nasal-only mask, but relies on a patient's own respiratory effort to determine the respiratory rate. Despite the fact that almost any mode applied via an ETT may be applied via NIV—including SIMV, SIMV/PS, PRVC, APRV, and others—current research into, and experience with, extubation and immediate application of NIV tends to utilize CPAP or CPAP/PS via face mask. The future may very well show that NIV via a nasal-only mask may be sufficient, facilitating feeding and oral care. Or perhaps NIV with the application of a mandatory rate may help achieve earlier extubation in all or a subset of patients. Furthermore, all of these modalities can also be applied to patients with a tracheostomy, allowing the patients to entirely determine their own TV and respiratory rate, but the concern is less for these patients who maintain a secured airway of a cuffed tube in the trachea.

In a review article, Epstein4 cites several uncontrolled studies of extubation directly to NIV. Many seem to show benefit but the studies suffer from several problems. First, none were RCTs. Second, many trials dealt with only a subset of patient pathology.5 While that is not an inherent problem, caution must be used when trying to generalize the results to all patients, or other subsets of patients. Of course, such results unmask benefits in certain subgroups—such as patients with COPD where there appears to be an exponential benefit—that might not be uncovered in larger studies with heterogeneous populations. Third, the specific NIV technique, as well as modality (face vs. nasal mask), was not standardized. Finally, each of the centers had experience in use of, and monitoring of patients on, NIV. These factors should be considered when applying the results of these studies to the extubation of a patient directly to NIV. Consider also that, while NIV has been growing in popularity in the ED, few emergency physicians apply NIV to the recently extubated patient. The initial settings and titrating of NIV as part of weaning program are not necessarily the same as those ordinarily used for the ED management of patients who receive NIV as part of their acute treatment for acute pulmonary edema or COPD.

Perhaps one of the most intriguing observations involves a rediscovery of conclusions drawn by Ely et al2 that part of the benefit of extubating patients and applying NIV may be artificially elevated by applying NIV to patients who might otherwise have been ready for extubation even without NIV.4,20 This correlates with several of Tobin's observations of the heuristics of extubation science.9 That effect, however, cannot account for all of the benefit. Additional benefit is also realized via the reduction of the risks of receiving mechanical ventilation: most notably ventilator-associated pneumonia. Furthermore, once an ETT is removed, there is no further indication for even mild amounts of sedation. Epstein concludes with an opinion of the criteria necessary to consider extubating a patient, earlier than might otherwise be considered, for immediate application of NIV4 (see Table 5-1).

Further recommendations for planned extubation to NIV are as follows:

1. The underlying condition necessitating the initial intubation should be resolved or, at least, resolving, with minimal remaining effect on respiratory status.

2. A large cuff leak with the balloon deflated, which is further described below.

3. Until further studied, a patient who has been extubated to NIV should remain in the ICU until no further assisted ventilation is required. In terms of acuity, monitoring, and management, a patient extubated to NIV still requires critical care.

In conclusion, the data do not, as of yet, support planning to extubate a patient and immediately placing him or her on NIV. There is, however, cause for optimism that this therapy may shorten the time patients spend sedated on a ventilator with an ETT and, perhaps, in the ICU overall. For now, though, this therapy should not be thought of as assisting an otherwise weaned patient, but rather as part of weaning. Patients receiving NIV after extubation should be monitored and receive critical care until they are fully weaned from the NIV. In fact, such patients may be even more tenuous than an otherwise similar patient who still has a cuffed tube in the trachea.

NIV as Treatment for Respiratory Insufficiency Following Extubation

As opposed to planning on applying NIV to patients immediately after extubation, consideration has been given to only applying NIV to patients who develop respiratory distress after extubation in an attempt to avoid reintubation. Unfortunately, there is significantly less investigation into this approach and the data have yielded conflicting results. In one of the earlier works, Keenan et al21 noted no benefit in the application of NIV when applied to patients only after they develop respiratory distress. One notable caveat is that after the first of their 3-year enrollment period, they stopped randomizing COPD patients since they believed, at that time, that the literature strongly favored extubating those patients to NIV.

Further confusing the data are findings perhaps best seen in the study by Nava et al: although NIV was associated in a 16% decreased risk for reintubation, those patients requiring reintubation had a 60% higher mortality risk.5 However, it is possible that confounding factors have obscured the benefits of extubation to NIV. First, subgroup analyses suggest the modality is beneficial to certain subgroups, such as those with CHF or COPD. Second, it may be that once patients are extubated to NIV, they may be thought to be more stable than they actually are, and may receive less therapy and monitoring than if they were intubated.

Therefore, as of now, the data do not support utilizing NIV for patients who develop respiratory distress after intubation. However, for the subset of patients with COPD, the data are not clear. The reasons for this are probably 2-fold. First, although not yet proven, patients with COPD benefit from NIV more so than other patients, so it is likely that some of these patients with postextubation respiratory distress will respond in the same manner as the tendencies shown thus far in COPD patients extubated directly to NIV. Second, because patients with COPD have shown a trend to decreased mortality and morbidity when extubated directly to NIV, many studies have either excluded COPD patients or been hesitant to randomize them due to ethical considerations. This means that current study results may not be necessarily applicable to COPD patients. Therefore, a large, prospective randomized controlled study is required to study the outcome of application of NIV to COPD patients who develop respiratory distress after extubation. For all other patients, however, the data support early reintubation of patients who develop respiratory distress after extubation.

Postextubation Stridor

A major cause of postextubation respiratory failure is upper airway pathology. Various studies have reported the frequency between 2% and 16%.22 Usually resulting from laryngeal edema, physiologic causes include local damage, release of inflammatory mediators, and systemic third spacing. This is demonstrated by studies showing that risk factors for postextubation stridor include severity of illness (as indicated by the Simplified Acute Physiology Score), medical admission, traumatic intubation, self-extubation, duration of intubation, balloon pressure of the ETT balloon,22 intubation without sedation or paralysis, neurologic deficit as determined by a decreased GCS, and female gender.23 Interestingly enough, with respect to “traumatic intubation,” no association has been found between the primary training or experience of the intubator (prehospital, resident, nurse anesthetist, nonanesthesia attending, or anesthesia attending) and the incidence of postextubation airway obstruction,23 further confirming that postextubation respiratory distress is a result of the chronic course and treatment of a patient's disease, rather than of any single, acute event.

Attention must then be turned to ways of predicting and treating postextubation airway obstruction. As discussed above, NIV has been shown to be an ineffective treatment. Although depending on how the data are examined, it is theoretically possible that an attempt does not confer additional harm. Overall, reintubation carries such high mortality and morbidity that it is better to not extubate these patients until postextubation airway problems will not occur.

A test, called cuff-leak test, has been postulated to accurately detect the incidence of postextubation respiratory distress. The concept of the test is straightforward: measure the TV with the ETT balloon inflated and deflated. The difference is the “cuff leak.” With the ETT balloon inflated, all air should exhale through the ETT, allowing measurement by the ventilator. With the ETT deflated, some air will go through the ETT and be measured by the ventilator. The rest will go around the ETT, or “leak” out, and this quantity will not be measured by the ventilator. A larger leak volume is good and indicates a less occluded or swollen airway.

Several methods for performing the cuff-leak test have emerged, but none have become standard in the literature. Measuring the nonleak volume is trivial; simply note the TV before deflating the ETT balloon. Many studies use the average over four to six breaths. Prior to measurement of the leak volume, it is prudent to suction the ETT as well as the pharynx and upper larynx. The balloon is then deflated, whereupon the patient will generally cough. After the patient ceases coughing, the measured TV can be recorded or, as above, the mean over four to six breaths can be calculated. At that point, the balloon is reinflated.

Originally the cuff leak was expressed as an absolute number of milliliters of air. However, in attempts to standardize reporting, and to make results generalizable to patients and ETTs of all sizes, Sandhu et al assert that cuff leak should always be expressed as a percentage of TV.24 To accomplish this, one simply divides the absolute cuff-leak volume by the TV obtained with the cuff inflated.

Sandhu does not consider cases where the cuff-leak value may be misleading. For patients with large TVs and large ETTs, a cuff leak might be small because the ETT takes up most of the airway, forcing all air through the ETT whether the cuff is inflated or not. This would result in a small leak in a patient whose airway is fine. Conversely, with an ETT smaller than that usually used for patients taking a given TV, there are lower limits below which even a large relative leak (i.e., a large percentage leak) would not negate the possibility of severe, immediate respiratory distress. This is because, in this case, the maximal cross-sectional area of a partially occluded trachea may still be larger than that of a small ETT and yet could still be too small for sufficient flow without the ETT. To date, there are no studies suggesting values for concern, so it falls to the provider to exercise due diligence without guidance from the literature. Furthermore, for naturally small airways (tracheal malacia, airway trauma or growths, prolonged or multiple prior intubations with possible scar tissue formation), it seems prudent to take additional precautions in the periextubation period with a low threshold for reintubation if the patient exhibits any signs of respiratory distress.

Given the various works in the area, and barring extremes of patient size compared with ETT size, it seems that a cuff leak of 10–12% is a reasonable cutoff for extubation. Patients with cuff leaks below 10–12% are at high risk for postextubation respiratory distress requiring reintubation.22–24

A Final Word About Cuff Leak

Prinianakis et al did an extensive and elegant study to further delineate the etiology of cuff leaks.25 They noted that when the TV for a cuff leak is measured, the ETT balloon is deflated and left deflated throughout the entire respiratory cycle, allowing air to “leak” during both the inspiratory and expiratory phases. Although leak during the expiratory phase is the desired quantity, what actually may be measured by this test is leak during both phases. This is because when the ventilator applies a breath with the ETT balloon deflated, some air will exit the ETT tube into the trachea and immediately go up, around the ETT, into the mouth, and out of the body. Yet the ventilator measurement will assume that the entire volume has been delivered to lungs. Prinianakis therefore asserts that the traditional approach overestimates the expiratory leak by a factor of two or more.

In order to isolate the inspiratory and expiratory components, during the “leak assessment” ventilations, Prinianakis et al deflated the ETT balloon only during the expiratory phase. However, to ensure isolation of the expiratory from inspiratory phase, patients were heavily sedated with propofol and fentanyl and then paralyzed with cis-atracurium. Furthermore, the TV was set at 10 mL/kg, which is much higher than the generally recommended 6–8 mL/kg. Finally, no PEEP was applied, again different from most patient treatment regimens. Under these parameters, patients were found to have a significantly higher cuff leak when the ETT balloon was deflated for the entire respiratory cycle when compared with deflation only during the expiratory cycle, suggesting a significant component of inspiratory phase leak, possibly obscuring the “true” expiratory cuff leak. Although the protocol used by Prinianakis should not be routinely performed in all patients, especially those in whom extubation is being planned, the study does suggest that the cuff-leak test as is usually performed is insufficient, and further study may yield a more accurate test providing different values.

Treatment Pre-Extubation and Postextubation

Whether a patient is suspected to have postexpiratory airway occlusion, or a cuff-leak test has demonstrated the possibility, attention must be given to treatments to prevent postextubation stridor. These treatments fall into two categories: pretreatment given pre-extubation to decrease the possibility of postextubation stridor and post-treatment given after extubation as prophylaxis or if stridor develops.

Pre-extubation treatment has generally focused on steroids. While studies have considered different steroids, most work has been done with either hydrocortisone or dexamethasone. However, as with other investigations into periextubation treatment, the regimens have not been standardized. One of the most successful studies by Cheng et al not only delineated cuff-leak thresholds predictive of postextubation stridor, but also established that the timing of IV steroids (>6 hours before extubation) is probably more important than the specific medication used or the number of doses given.

Other treatment modalities such as diuretic therapy for 24 hours before extubation, inhaled racemic epinephrine, heliox, and anesthetic gases have all been shown to contribute to the reduction of postextubation stridor. The literature has yet to delineate which treatments are the most beneficial and in exactly which patient populations but it is important for the practitioner to keep these modalities in mind when considering extubating a patient with a concerning cuff-leak test. Currently the data show a single dose of 40 mg of methylprednisolone IV 24 hours prior to extubation reduces the incidence of postextubation stridor.

The Physical Act of Extubation

Once the patient has been weaned, consideration may be turned to removing the artificial airway. Unfortunately, the act itself is an unpleasant one for the patient. Care must be taken in the process, not only to reduce patient discomfort but also because the procedure itself as well as the patient's involuntary reactions, such as gagging, may be harmful.

Before beginning, all necessary equipment should be assembled (see Figures 5-3 and 5-4), including postextubation treatment modalities such as humidified oxygen. Note that the equipment list includes reintubation equipment and, depending on the patient, may also include difficult airway devices as well. Although not pictured, some practitioners like to place a folded-up sheet, absorbent pad, or water-impermeable barrier (such as a blue underpad or “Chux”) on the patient's chest to catch any secretions. This also provides a convenient location to quickly place the ETT once removed (while keeping the patient clean) so attention can be immediately turned to the rest of the procedure.

Since secretions pool in the posterior oropharynx, the oral cavity should be suctioned down to the posterior oropharynx using a soft suction device prior to extubation. Furthermore, secretions also pool in the ETT so that, too, should be suctioned. While preoxygenating a patient prior to suctioning is widely practiced, some authorities also advocate preoxygenating with 100% oxygen for 1–2 minutes prior to extubation.

When ready to extubate, the procedure involves deflating the ETT balloon and removing the ETT. Some advocate actually cutting the pitot tube so the balloon will deflate on its own. However, the balloon may not empty entirely, and withdrawing it through the vocal cords may damage or inflame the cords, causing damage or swelling. If the pitot tube is to be cut, it is reasonable to first use a syringe to deflate the balloon as much as possible.

Once the balloon is deflated, patients will usually start coughing for a variety of reasons. First, the pressure sensation in the trachea changes, which can stimulate coughing. Second, airflow changes when the balloon is deflated that may make it harder for patients to move air. Third, secretions in the trachea from above the balloon or in the posterior oropharynx may drip down into the patient's airway, causing a cough. Finally, movement of the ETT as it is extracted presses on the throat and can, itself, directly cause a cough.

To actually remove the tube, have the patient take a deep breath, deflate the ETT balloon, and have the patient forcefully exhale, removing the ETT as the patient exhales. Having the patient exhale prevents aspiration and reduces the likelihood of causing vocal cord damage by removing the tube during maximal vocal cord abduction.

For patients who have concomitant orogastric, nasogastric, or jejunal tubes, those tubes may be removed simultaneously with the ETT. For gastric tubes, many practitioners place them on suction for some time before extubation to prevent aspiration. Regardless, make sure the tube is not to suction during the removal or extubation.

Once the ETT has been removed, suction the oropharynx for any secretions dislodged by removal of the ETT and apply humidified oxygen (Figure 5-5). Use cool mist since heated mist may increase swelling. Note that the patient may require a higher FiO2 than just before extubation. Humidified systems allow almost any FiO2 to be set but 40% is a typical value used (Figure 5-6). If indicated, consider nebulized racemic epinephrine to combat postextubation edema.

Although humidified oxygen is most commonly applied using an aerosol mask, some practitioners prefer to place a face tent (Figure 5-4, item #13). Furthermore, some patients find the aerosol mask uncomfortable but tolerate a face tent better. Finally, for patients with a history of facial trauma, an aerosol mask may be contraindicated in which case a face tent should be used.

It is important to reassess the patient after extubation. In particular, check for stridor, bilateral and equal breath sounds, and significant changes in vital signs. Note that because of the noxious nature of the procedure, a slight elevation in pulse and blood pressure is common and usually will resolve without treatment. Extreme changes in vital signs should prompt a full reassessment of the patient.

Complications of Extubation

There are very few immediate complications of extubation. When present, however, they can be devastating.

Laryngeal spasm, though rare, can be an immediately life-threatening complication of extubation. While most laryngospasms spontaneously subside within seconds, any laryngospasm lasting longer than a few seconds demands immediate action. Reintubation using neuromuscular blockade will almost certainly be required, with simultaneous preparations for an emergency cricothyrotomy if the spasm prevents intubation through the vocal cords.

Vocal cord edema from irritation by ETT removal may be immediate or delayed. Ensuring that the ETT balloon is fully deflated should make this an infrequent occurrence. If it does occur, adjunct modalities such as humidified oxygen, racemic epinephrine, or even NIV may help. If not, the airway is in danger and reintubation or even cricothyrotomy might be required.

The most common complication is respiratory distress. This may resolve or may necessitate reintubation. Postextubation respiratory distress may be treated with racemic epinephrine and/or NIV as discussed earlier. Unfortunately, most patients requiring reintubation are fine in the periextubation period, but fail minutes to hours later. This indicates that recently extubated patients must continue to be monitored in an appropriate setting.

A Final Word on Predicting Successful Extubation

The best parameter currently known to predict successful extubation is the RSBI. But given its imperfect specificity, and the desire to, as soon as possible, extubate patients that will or even may tolerate extubation, the most important safety parameter may be the intensivist's comfort with dealing with the complications of failure: specifically, the need for reintubation.

This places the emergency physician in the position of being among the best specialties to practice as intensivists. Although practitioners of other specialties can pursue and master the art of intubation, except for anesthesiology, no other specialty encompasses—and trains for—the need for emergent intubation like emergency medicine. In the emergency department, patients who require airway management require intubation regardless of airway anatomy, last meal eaten, or any of the other factors controlled for with elective procedures. It is perhaps this mindset that stands practitioners of emergency medicine in good stead.

Since extubating a patient carries a risk of reintubation, the best plan is to perform a 30-minute SBT and a cuff-leak test. If the patient passes both, extubation may proceed. Obviously, the practitioner must always be ready to reintubate should the patient decompensate clinically.

Other considerations involve the resources necessary to reintubate a patient. Since staffing levels vary with time of day, it seems reasonable to limit elective extubations to those times when maximal resources are available. For instance, one should consider not extubating patients at night.

Furthermore, for patients known or suspected to be a difficult intubation, it would seem prudent to make sure maximal resources are immediately available prior to extubating the patient. This may mean having fiberoptic or other rescue equipment at the bedside or perhaps planning for extubation in the OR with a double setup for cricothyroidotomy. Some patients for whom this may be required include known difficulty when originally intubated, history of airway trauma, airway masses, prior tracheostomy, or other airway surgical procedures.

To wean patients on mechanical ventilation, the following points would seem to be supported:

• Weaning should take place as soon as possible; in many cases, weaning can and should be started immediately on initiation of mechanical ventilation.
• FiO2 may be titrated independently of ventilation, using pulse oximetry.
• Protocols should be instituted so that once- or twice-daily SBTs are performed automatically by the respiratory therapist or critical care nurse with calculation of the RSBI.
• Modes that allow patient to control minute ventilation are superior. SIMV is preferable to CMV since CMV requires the patient who initiates an extra breath to take the full programmed TV.
• Modes should always be applied with PS or ATC so that additional, patient-initiated breaths are provided at a level comparable to what a patient would experience if not intubated.
• Modes, including TV and rate, should be titrated down based on patient comfort, allowing the patient to “take over” more and more of his or her own respiratory effort.
• For alternate ventilatory modes, weaning should proceed in a manner best suited to that mode, with extubation directly from that mode or conversion to CPAP/PS and progressed as above. For specific weaning parameters for each mode (SIMV/PS, APRV), see above.
• When at a minimum level of support, the ventilatory mode should be changed to CPAP/PS with levels similar to those provided to the patient while on SIMV/PS.
• When the patient is at CPAP level of 5, or a CPAP/PS level of 5/10, a RSBI should be calculated for 30 minutes.
• If the RSBI is ≤105, the patient should be considered for extubation and cuff-leak test performed. If the patient has a tracheostomy and has a RSBI ≤105, the patient should be considered for removal from the ventilator to be placed on a tracheostomy collar.
• For those with laryngeal swelling as evidenced by a concerning cuff-leak test, consider pre-extubation steroids, diuretic therapy, and/or NIV.
• For those with RSBI ≤105 and a cuff leak >10–12%, consider extubation as soon as is practical and safe.

The author would like to extend his gratitude to Reneé Rainey, RRT, RN, for assistance with the equipment pictured in Figures 5-1 to 5-6 and to Zev Perlmutter for technical assistance and postprocessing of Figures 5-1 and 5-4.