Why Might It Be Difficult to Withdraw Patients From the Ventilator?

In concept, there are 2 fundamental reasons that patients are unable to be withdrawn from mechanical ventilator support.¹⁶ The first is that they are “too sick.” This is illustrated in Figure 1,¹⁷ which defines ventilator dependence as being caused by an imbalance between demands and capabilities. High demands come from high ventilation/oxygenation requirements and/or abnormal respiratory system mechanics. Impaired capabilities come from poor neural ventilatory control and/or compromised respiratory muscle function. Depending upon the disease state and underlying health status, this imbalance can resolve quickly (eg, drug overdose) or can require weeks to months to recover. Indeed, in some patients the imbalance is never restored and life-long mechanical ventilation is required.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

Schematic representation of the balance between patient demands and capabilities. When demands outweigh capabilities, mechanical ventilation is needed. When capabilities outweigh demands, ventilator discontinuation is possible. (From Reference 17).

The second fundamental reason that patients are unable to be withdrawn is clinician behavior.¹⁶ Large clinical trials from the 1990s clearly showed that clinicians frequently did not appreciate in a timely fashion that ventilator withdrawal was possible. This was often made worse by slow “weaning” procedures that actual delayed the withdrawal process.^13,14 Clinicians can also adversely affect the demand/capability balance in Figure 1 by producing ventilator-induced lung injury from non-lung-protective settings,¹⁸ by increasing imposed loads from asynchronous ventilator settings,¹⁹ or by inappropriately blunting neural drive (eg, sedatives/paralytics)²⁰ or weakening muscles (eg, toxins, poor nutrition).

Scope of the Problem

A large observational study from Europe involved 4,559 mechanically ventilated patients from 349 ICUs in 2004.²¹ These data were analyzed using a newly described classification system of ventilator discontinuation²²: “Simple” (ventilator discontinued after the first assessment), “Difficult” (ventilator discontinued from 2–7 d after initial assessment), and “Prolonged” (ventilator discontinued in > 7 d after initial assessment). In this analysis, 2,714 patients were successfully discontinued over the 28 day study period.²³ Fifty-five percent were simple discontinuation, 39% were difficult discontinuation, and 6% were prolonged discontinuation. Not surprisingly, patients needing prolonged discontinuation processes were sicker and had longer stay and higher mortality than the simple discontinuation patients.

Patients requiring mechanical ventilation beyond the 28 day mark are often considered to be requiring prolonged mechanical ventilation (PMV). The PMV population is growing rapidly and reflects a number of factors, including the aging of the general population and the ability of the healthcare system to keep patients alive longer after devastating illnesses or aggressive surgical procedures.^24–26 PMV patients are estimated to be as many as 5–10% of all mechanically ventilated patients in the United States. While in-patient mortality in PMV patients may be as high as 35%, as many as half of the survivors will be successfully withdrawn from mechanical ventilator support, usually within the first 90 days.²⁴ Thereafter, the likelihood of successful ventilator withdrawal is very low.

The Evidence Base Supporting Ventilator Withdrawal Strategies in the Intensive Care Setting

The important clinical questions facing the clinician in the ICU are: When can efforts to discontinue ventilation be initiated? What assessment strategies will best identify the patient who is ready for ventilator discontinuation? When should extubation be carried out?

Evidence to answer these questions comes largely from observational studies in which a certain parameter (or set of parameters) is compared in a group of patients who either successfully or unsuccessfully have been removed from the ventilator. The general goal of these studies is to find “predictors” of outcome. Evaluating results from these types of studies can be difficult for several reasons.²⁷ First, the “aggressiveness” of the clinician/investigator's discontinuation philosophy needs to be understood, as it will affect the performance of a given predictor.

Second, patients are recruited into these studies because investigators believe there is some reasonable chance of success for ventilator discontinuation. These “entry” criteria often include some form of clinical judgment or intuition, making results from one study difficult to compare to another.

Third, methodological problems inherent to observational studies include different measurement techniques of a given parameter from study to study, large coefficients of variation of a given parameter with repeated measurements from study to study, and different patient populations (eg, long-term vs short-term ventilator dependence).

Fourth, assessed outcomes differ from study to study. Some investigators have examined successful tolerance of a spontaneous breathing trial (SBT), others have used permanent discontinuation of the ventilator, and others have combined successful discontinuation and extubation. In addition, different studies use different durations of ventilator discontinuation or extubation to define success or failure. Although 24–48 h of unassisted breathing often is considered to define the successful discontinuation of ventilator support in the ICU setting, many studies use shorter time periods to indicate success and often do not report subsequent reintubation rates or the need to reinstitute mechanical ventilatory support.

In 1999, McMaster University, funded by a large grant from the United States Agency for Health Care Policy Research, published a comprehensive evidence-based review of the world literature pertaining to ventilator discontinuation (over 5,000 publications).²⁸ This report found evidence in the literature supporting a possible role for 66 specific measurements as predictors of successful ventilator discontinuation. To evaluate the role of these parameters, the McMaster report used likelihood ratios (LRs), an expression of the odds that a given test result will be present in a patient with a given condition, compared to a patient without the condition. An LR > 1 indicates that the probability of success increases, while a value < 1 indicates that the probability of failure increases. LRs between 0.5 and 2 indicate that a discontinuation parameter is associated with only small, clinically unimportant changes in the post-test probability of success or failure. In contrast, LRs of > 10 or < 0.1 correlate with very large changes in probability. With this approach, the McMaster group identified 7 parameters that had consistently significant LRs to predict successful ventilator discontinuation in several studies. Some of these measurements are made while the patient is still receiving ventilatory support; others require an assessment during a brief period of spontaneous breathing. These parameters, their threshold values, and the range of reported LRs are given in the Table. It should be noted that, despite the statistical significance of these parameters, the generally low LRs indicate that the clinical applicability of these parameters alone to individual patients is low.

Following the publication of the McMaster report, the American College of Chest Physicians, the Society of Critical Care Medicine, and the American Association for Respiratory Care (ACCP/SCCM/AARC) assembled a task force to issue evidence-based guidelines for clinicians to follow in the ventilator discontinuation process.¹ These guidelines were based largely on the McMaster review of the clinical evidence base, but by necessity also incorporated evidence from basic science work, lung model studies, animal studies, non-outcome-based human studies, and even expert opinion to “fill in the gaps” in the clinical evidence base. In the end, 12 guidelines were described and published.¹

According to these guidelines, the first step in the discontinuation process is to regularly assess the status and trajectory of underlying cause(s) for the need for mechanical ventilatory support in a given patient.¹ Among the more important factors impacting ventilator dependence are neurologic abnormalities affecting the brainstem ventilatory control system, respiratory muscle capability/mechanical load imbalances, impaired ventilation/perfusion matching in the lungs, abnormal cardiac function, lung edema, metabolic derangements (eg, glucose homeostasis and adrenal function), ultimate oxygen delivery, and even psychological factors. A focused search for the underlying causes for ventilator dependence may be especially important in “difficult to discontinue” patients, as previously unrecognized but reversible conditions may be discovered.²⁹ As noted above, iatrogenic factors such as excessive sedation use, inappropriate ventilator settings causing lung injury and/or discomfort, inappropriate fluid management, inadequate nutrition, and lack of patient physical activity may also contribute to ventilator dependence. And, of course, a failure to recognize ventilator discontinuation potential will also lead to iatrogenic ventilator dependence.

The criteria used by clinicians to define disease “reversal,” however, have been neither defined nor prospectively evaluated in a randomized controlled trial. Rather, various combinations of subjective assessment and objective criteria (eg, usually gas exchange improvement, mental status improvement, neuromuscular function assessments, and radiographic signs) that may serve as surrogate markers of recovery have been employed.^{3,9–11,13,14,30} It should be noted, however, that some patients who have never met one or more of these criteria still have been shown to be capable of eventual liberation from the ventilator.¹¹

Although there needs to be some evidence of “clinical” stability/reversal, a more focused assessment is needed before deciding to continue or discontinue ventilatory support. The ACCP/SCCM/AARC guidelines¹ state that this formal assessment should be an SBT. This is based on the very strong evidence that, although assessments that are performed while a patient is receiving substantial ventilatory support or during a brief period of spontaneous breathing can yield important information about discontinuation potential (LRs in the Table), integrated assessments that are performed during a formal, carefully monitored SBT appear to provide the most useful information to guide clinical decision making regarding discontinuation.¹ In concept, the SBT should be expected to perform well, as it is the most direct way to assess a patient's performance without ventilatory support. Multiple studies have found that patients tolerant of SBTs were found to have successful discontinuations at least 77% of the time.^{1,3,9,13,15,29,31} However, because patients failing the SBTs in these studies were not systematically removed from ventilatory support, the ability of a failed SBT to predict the need for ventilator dependence (ie, negative predictive value) cannot be formally assessed. Indeed, it is conceivable that iatrogenic factors such as endotracheal tube discomfort or demand-valve insensitivity/unresponsiveness, rather than true ventilator dependence, caused the failure of the SBT in at least some of these patients.^3,15,32–35 Thus, it is unclear how many patients who are unable to tolerate an SBT would still be able to tolerate long-term ventilator discontinuation. Although the number is likely to be small, it is probably not zero, and this needs to be considered when dealing with patients who repeatedly fail an SBT.

The criteria used to define SBT “tolerance” should be integrated indexes, since, as noted above, single parameters alone perform so poorly. These integrated indexes usually include several physiologic parameters (eg, gas exchange, ventilator pattern, hemodynamics) as well as clinical judgment incorporating such difficult-to-quantify factors as “anxiety,” “discomfort,” and “clinical appearance.” Importantly, these parameters need to be assessed in the context of each other and as changes from baseline, not as rigid thresholds. Indeed, one important recent study addressing this issue showed clearly that the use of a rigid threshold of the ratio of breathing frequency to tidal volume < 105 to define SBT success in fact slowed the discontinuation process.³³ This observation has implications both for written management protocols as well as computer driven protocols.

The evidence is strong supporting the recommendation that an SBT should be at least 30 min but no longer than 120 min to allow proper assessment of ventilator discontinuation potential.¹ This means that clinicians should wait at least 30 min to assure SBT tolerance, but terminate the trial at 120 min if SBT tolerance is still unclear. There is evidence that the detrimental effects of ventilator muscle overload, if they occur, often occur early in the SBT.^15,34–36 Thus, the initial few minutes of an SBT should be monitored closely before a decision is made to continue (this is sometimes referred to as the “screening” phase of an SBT).

Controversy exists on the “best” technique used to do the SBT. Options include a simple T-piece, where only supplemental O₂ is supplied at the proximal end of the endotracheal tube, setting the ventilator to a CPAP level equivalent to the previous PEEP setting, or setting a low level of assistance (eg, pressure support of 5–8 cm H₂O or the use of “automatic” tube or airway compensation). The T-piece approach comes closest to mimicking the situation the patient will experience when extubated, and it is recommended by some for maximizing specificity of the test (ie, lowest number of false positives). However, because the endotracheal tube is still present with its associated discomfort, the sensitivity of the SBT (ie, detecting true positives) may be somewhat compromised. In contrast, supplying low levels of inspiratory and/or expiratory pressure may hide a patient's inability to tolerate complete ventilator removal (excessive false positive tests), although it may relieve some of the iatrogenic discomfort of the endotracheal tube (fewer false negative tests). In large population studies, all 3 approaches appeared to perform well, but the T-piece approach might be considered if there is concern about borderline SBT performance with other techniques or there is concern about the potential effects from a loss of PEEP.^9,36–40

A potential concern about the SBT is safety. Although unnecessary prolongation of a failing SBT conceivably could precipitate muscle fatigue, hemodynamic instability, discomfort, or worsened gas exchange,^41–45 there are really no data showing that SBTs contribute to any adverse outcomes if terminated promptly when failure is recognized. Indeed, in a cohort of > 1,000 patients in whom SBTs were routinely administered and properly monitored as part of a protocol, only one adverse event was thought to be even possibly associated with the SBT.¹¹

While ICU patients who are judged tolerant of the SBT should move on to assessments of the need for continued use of the artificial airway (see below), the patient who is judged not tolerant of the SBT requires a different approach. In these patients there are 3 specific issues to address.¹ First, a careful search once again should be undertaken for ongoing (and potentially reversible) causes of ventilatory dependence. Second, a comfortable interactive form of ventilatory support should be provided that encourages respiratory muscle activity but does not overload muscles or compromise gas exchange.¹⁹ Importantly, there are few (if any) data demonstrating that attempts to “wean” this support, rather than keep it constant, are beneficial.¹⁶ Third, every 24 hours the patient should be reassessed for another SBT. Importantly, all of these procedures can be carried out through protocols run by skilled clinicians (eg, respiratory therapists).^1,11,30 Indeed, protocols have been shown in multiple studies to facilitate the ventilator withdrawal process and shorten the duration of mechanical ventilation.^46,47

Since the publication of the original ACCP/SCCM/AARC guidelines,¹ 2 important developments have occurred that build on the SBT approach. The first has been the application of computer driven assessments and clinician feedback tools that adjust support and remind clinicians when SBTs are needed.^48,49 These tools are discussed in more detail below, but their most important feature may be simply to remind clinicians to perform regular SBTs.¹⁶ Second, and perhaps more importantly, has been the linkage of the SBT strategy to a sedation optimization strategy.^50,51 As noted above, excessive sedation use has been recognized for years to be a barrier to effective SBT performance and efficient ventilator withdrawal. Optimizing patient-ventilator synchrony with appropriate ventilator settings can help minimize sedation use, but several recent studies have also emphasized that focused protocols aimed at reducing sedation usage further can add to this. Examples of such protocols link the SBT to either spontaneous awakening trials (the routine cessation of all sedatives)⁵⁰ or to focused protocols aimed at minimizing sedation.⁵¹ An example of a combined SBT and spontaneous awakening trial protocol is given in Figure 2. Regardless of the sedation minimization strategy used, this linkage to routine SBTs has been shown to markedly accelerate the ventilator withdrawal process.^50,51

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

A conceptual “backbone” to a combined spontaneous awakening trial and spontaneous breathing trial ventilator discontinuation protocol. (From Reference 16.)

It has been over a decade since this comprehensive set of evidence-based guidelines for the ventilator discontinuation process was issued.¹ The most interesting assessments of the early impact of these guidelines were 2 re-analyses of a large observational database noted above, which involved 4,559 mechanically ventilated patients from 349 ICUs in 2004.²¹ In the first re-analysis⁵² the use of SBTs as the first discontinuation assessment technique was found to have increased from 58% to 62% from a similar survey in 1998 (P = .09). In the second re-analysis,⁵² discontinuation strategies using gradual support reductions (synchronized intermittent mandatory ventilation [SIMV] or SIMV plus pressure support) were reduced significantly, from 11% and 26%, respectively, in 1998, to 1.6% and 15%, respectively, in 2004 (P < .001). Importantly, time devoted to the discontinuation process decreased from 50% to 40% of total ventilation time over the same period (P < .001).

Taken together, these data suggest that clinical use of SBTs is commonplace and gradually increasing, especially in patients judged to be clinically ready for discontinuation. However, there still appears to be a persistent aversion to SBTs in the majority of patients about whom clinicians have concerns. This may not be optimal and, indeed, the evidence would suggest that patients should not be labeled difficult or prolonged discontinuation problems until they have failed multiple SBTs.

The Evidence Base Supporting Ventilator Withdrawal Strategies in the PMV Patient

The PMV patient is generally a patient who has already failed multiple SBTs and thereby has declared him/herself to have a very slowly resolving disease process.^24–26 Under these conditions, repeated daily SBTs are likely futile until substantial disease resolution has occurred and the balance in Figure 1 is clearly improving. In the PMV patient, gradual ventilator support reduction strategies (“weaning”) now may make some sense. Importantly, these support reduction strategies are not therapeutic: they are not designed to improve the load/capabilities balance of Figure 1 in and of themselves. Instead, tolerance of support reduction strategies would be used to assess disease resolution to determine when it is appropriate to resume SBTs. Although no clinical trials exist in this arena, most authorities would recommend that tolerance of a reduction of support to 50% of the original level is a signal to restart SBTs.²⁴

Because most PMV patients have tracheostomies in place, the SBT strategy is often different from the acute care setting. Indeed, it is usually not a 30–120 min formal trial, but, rather, a gradually increasing duration of tracheostomy collar breathing. The initial trials are often short and are subsequently lengthened until the patient tolerates being off the ventilator during waking hours. Thereafter, attempts at unassisted tracheostomy collar breathing through the night can be attempted.

Can the Ventilator Withdrawal Process Be Automated?

In recent years a number of feedback systems have been introduced that are designed to automatically reduce the level of ventilator support in patients recovering from respiratory failure.^15,48,49 These modes usually provide pressure support breaths, and the applied pressure is reduced automatically. Different algorithms are available, and include simple tidal volume targets, tidal volume targets with rate and inspiratory/expiratory time considerations, and tidal volume targets with rate and end-tidal CO₂ considerations.

Importantly, with all of these support reduction strategies the assumption is that support reduction is useful before performing SBTs—an assumption that has little supporting evidence in the ICU.¹⁶ Nevertheless, clinical evidence exists that these modes function as designed and that in settings with rapidly recovering patients and low clinical staffing they may help alert clinicians that SBTs are indicated. Good evidence showing better outcomes with these approaches, compared to proper evidence-based withdrawal strategies in more general ICU settings, does not exist.

Clinicians using these automated support reduction strategies need to understand them well. This is particularly true if using systems based only on a tidal volume target. With this approach, an anxious or uncomfortable patient may generate enough tidal volume that the needed pressure support is actually reduced inappropriately.⁵³ If these automated systems have any utility, it would most likely be either in the setting of a rapidly recovering patient in an ICU or perhaps during the gradual recovery process of the PMV patient, as described above.

Assessing the Need for an Artificial Airway

Once a patient has been deemed to no longer need mechanical ventilatory support (or perhaps is deemed a candidate for noninvasive ventilation),^42,54 attention then turns to the assessment of the need for the artificial airway. Extubation failure can occur for reasons distinct from those that cause discontinuation failure. Examples include upper-airway obstruction or the inability to protect the airway and to clear secretions.

The risk of post-extubation upper-airway obstruction increases with the duration of mechanical ventilation, female sex, trauma, and repeated or traumatic intubation.^3,55 The detection of an air leak during mechanical ventilation when the endotracheal tube balloon is deflated can be used to assess the patency of the upper airway (cuff leak test).^55,56 In a study of medical patients, a cuff leak of < 110 mL (ie, average of 3 values on 6 consecutive breaths) measured during continuous mandatory ventilation within 24 h of extubation identified patients at high risk for post-extubation stridor.⁵⁷ Although others have not confirmed the utility of the cuff leak test for predicting post-extubation stridor,⁵⁸ many patients who develop this can be treated with steroids and/or epinephrine (and possibly with noninvasive ventilation and/or heliox) and do not necessarily need to be reintubated. Steroids and/or epinephrine also could be used 24 h prior to extubation in patients with low cuff leak values or who are otherwise considered at high risk.⁵⁹ It is also important to note that a low value for cuff leak may actually be due to encrusted secretions around the tube rather than to a narrowed upper airway.

The capacity to protect the airway and to expel secretions with an effective cough would seem to be vital for extubation success, although specific data supporting this concept are few. Airway assessments generally include noting the quality of cough with airway suctioning, the absence of “excessive” secretions, or the frequency of airway suctioning (eg, every 2 h or more).^3,60,61 One approach uses an “airway care score,” which semi-quantitatively assesses cough, gag, suctioning frequency, and sputum quantity, viscosity, and character, that predicts extubation outcomes.⁴ Peak cough flows of > 160 L/min predict successful translaryngeal extubation or tracheostomy tube decannulation in neuromuscular or spinal cord-injured patients.⁶² Cough velocities of 0.5–1.0 L/s have also been shown in other studies to be compatible with successful extubation.⁶³

The importance of intact cognitive function on extubation success is controversial. Successful extubations have been reported in a select group of brain-injured, comatose patients who were judged to be capable of protecting their airways.⁴ However, it is difficult to extrapolate this experience to more typical ICU patients, and many would argue that some capability of the patient to interact with the care team should be present before the removal of an artificial airway. Nevertheless, a review of the literature suggests that a Glasgow coma score above 8 is compatible with successful extubation, provided that adequate airway protection capabilities exist.⁶⁴

The Role of Tracheostomies

A tracheostomy offers several putative advantages over a translaryngeal airway.⁶⁵ These include better patient comfort (and thus less need for sedation), ability to eat, and less infection risk. However, a tracheostomy is an invasive procedure with complications of long-term tracheal injury. In the patient who is not a simple ventilator withdrawal candidate (see above), the timing of tracheostomy placement is controversial. Proponents of “early” tracheostomy argue that in a patient judged likely to need an artificial airway for more than 21 days, the benefits of early tracheostomy outweigh any risks and justify the costs. These clinicians would thus place the tracheostomy as soon as this judgment is made. Proponents of “late” tracheostomy argue that it is difficult to predict who will need a ventilator for more than 21 days until the patient is close to the 21 day mark. Waiting until then avoids unnecessary tracheostomies.

One of the most supportive studies for “early” tracheostomies randomized 120 patients who were judged likely to need ventilator support for at least 16 days to either get an immediate or delayed tracheostomy (ie, at day 16–18).⁶⁶ Interestingly, the clinicians were quite accurate in predicting this need, as 83% of the patients randomized to late tracheostomy actually got one. The important outcomes showed a significant mortality benefit, pneumonia benefit, and ventilator stay benefit to early tracheostomy (32% vs 62%, 5% vs 25%, and 7.6 d vs 17.4 d for early vs late, respectively).

A recent Cochrane systematic review analyzed this controversy and included 5 randomized trials (including the one above).^67,68 Their conclusion was that, despite considerable methodologic heterogeneity, an early tracheostomy strategy was associated with trends in favor of early tracheostomy in mortality (49% vs 64%) and pneumonia (12% vs 22%), as well as a significant reduction in time on ventilator in favor of early tracheostomy (7 d vs 17 d).

Summary

The ventilator discontinuation process is an essential component of overall ventilator management. Continued ventilator dependence can be a result of persistent illness or can be a result of poor management, and it is obviously important for the clinician to be able to assess both of these. An evidence-based task force has developed 12 evidence-based guidelines to help with these processes. Critical among these is the recommendation for regular SBTs. More recent developments have focused on the importance of linking sedation reduction protocols to ventilator discontinuation protocols. Patients with repeated SBT failures are often considered to require PMV. These patients often receive tracheostomies and are probably better managed with more gradual reductions in support and gradually lengthening spontaneous breathing periods. The evidence base is growing, but the earlier guidelines are standing the test of time. Moreover, practice patterns are evolving in accordance with them. Nevertheless there is still room for improvement and still the need for further clinical studies.