Prior to extubation, the cuff leak is usually checked. This consists of deflating the cuff of the endotracheal tube to verify that gas is able to move around the tube. Absence of a cuff leak suggests the presence of airway edema, increasing the risks of post-extubation stridor and reintubation. However, the test isn't that great. Absence of cuff leak can also occur due to having a large tube relative to the patient's trachea, or caked secretions around the tube. Thus, many patients without a cuff leak may still be safely extubated.
Uncertainty spawns practice variation. I've encountered attendings who span the entire gamut between two extremes:
- Aggressive approach: If there is no cuff leak, just ignore it. Extubate the patient and re-intubate as needed.
- Conservative approach: If there is no cuff leak, start steroids and leave the patient on ventilation. The next day, evaluate for a cuff leak. Delay extubation for 24-48 hours while attempting to elicit a cuff leak.
A new joint practice guideline by the ATS and ACCP addresses this issue. This guideline recommends a clever compromise between these extremes, which is the basis of the algorithm below. This provides a streamlined, evidence-based pathway to extubate patients without a cuff leak.
The absence of a cuff leak increases the likelihood of reintubation (positive likelihood ratio ~4) whereas the presence of a cuff leak reduces the likelihood of reintubation (negative likelihood ratio ~0.5)(Girard 2017). This affects the post-test probability of reintubation as follows (1):
The ideal extubation failure rate is debatable, with most sources suggesting ~15% as reasonable(2). Among patients with a low pre-test probability of reintubation (0-5%), even if there is no cuff leak then the risk of reintubation is still acceptable. Such patients may be extubated regardless of test result. Thus, among patients with a low pre-test probability of reintubation, there is no utility in performing this diagnostic test.
What if the patient has a low risk of reintubation, but the cuff leak test is checked anyway and is absent? The same statistical considerations still apply. Therefore, you are free to proceed immediately to extubation if you wish to do so.
Repeat the cuff leak test
This isn't addressed in the guidelines, but still merits discussion. There is no uniform definition of what constitutes a cuff leak. The following are some commonly used criteria:
- Audible leak
- Volume loss around cuff >110 ml (difference between inspired/exhaled volumes)
- Volume loss around cuff >24% tidal volume
To make matters even worse, test performance depends on ventilator settings. For example, old studies utilizing the >110 ml criteria used fairly high tidal volumes (e.g. 10 cc/kg; Lee 2007). This test cutoff may not work when using lower tidal volumes.
My practice is simply to listen for an audible leak, because this is easy and perhaps more accurate (Schnell 2017 found it to be slightly better than using a leak volume >110 ml). However, a falsely absent cuff leak may occur if the peak pressure is low.
Ultimately, there are numerous ways to test for cuff leak. However, any method should be used consistently, with careful attention to how tidal volume and airway pressure affect the results. Thus, a report that the patient has no cuff leak should be re-examined at the bedside, rather than accepted on face value.
For high-risk patients with no cuff leak, give steroid and extubate after >4 hours
This is the crux of this guideline, because it creates a pathway to extubate most patients during the same day. Spontaneous breathing trials are generally done early in the morning, so there is enough time to give steroid and extubate the patient in the early afternoon.
Steroids work rapidly, so there is no need to wait 24 hours
Traditional management of the patient without a cuff leak often involved giving steroids and re-evaluating for extubation the next day. However, prospective double-blind RCTs suggest that steroid causes rapid improvement in edema, with a maximal benefit within 4-6 hours. This indicates that it's unnecessary to wait a full day for steroid to work.
Lee 2007 performed a double-blind RCT of 86 patients who were intubated >48 hours with a cuff leak volume <110 ml. Patients were randomized to placebo vs. dexamethasone 5 mg IV Q6hr for 24 hours. Cuff leak volumes were checked periodically in both groups. Steroid caused substantial improvement in the first six hours, with little improvement thereafter:
Cheng 2011 performed a double-blind RCT of 71 patients with a cuff leak <24% of the tidal volume. Patients were randomized to a single dose of placebo versus 40 mg methylprednisolone, four hours prior to extubation. Steroid improved the cuff leak rapidly (figure below), decreased the rate of post-extubation stridor (16% versus 40%; p<0.05), and decreased the rate of reintubation (8% versus 30%; p<0.05). This study proves that clinically meaningful improvements may be obtained by giving steroid only four hours prior to extubation.
Like most steroid doses in critical care, nobody really knows. Commonly used regimens include dexamethasone 4 mg IV Q6hr or methylprednisolone 40 mg IV Q6hr for 24 hours.
Post-extubation laryngeal edema is most problematic within the first few hours after extubation. Therefore, it makes little sense to use a prolonged steroid regimen (e.g. dexamethasone 4 mg IV Q6hr). Dexamethasone has a long biological half-life (~48hr), so this regimen will actually cause dose-stacking with an increase in the steroid level over time. This produces maximal steroid effects when the steroid dose is stopped at the end of the course, a time when laryngeal edema is least problematic.
Given the timing of laryngeal edema, it makes the most sense to use a front-loaded steroid regimen (e.g. a one-time dose of 60 mg IV methylprednisolone). This would achieve a high steroid level rapidly, at the time points when laryngeal edema is most problematic. Since methylprednisolone has a biological half-life of 24 hours, it may not need to be re-dosed at all. Cheng 2006 showed that a single dose of 40 mg methylprednisolone had equal efficacy compared to 40 mg IV Q6hr for four doses (and later validated the practice of giving a single dose of methylprednisolone in their 2011 publication).
After steroid therapy, extubate regardless of the presence/absence of cuff leak
After you've given steroid and waited >4 hours, the guidelines recommend extubation even if there is still no cuff leak. Although counterintuitive, this is justified for several reasons:
- Cuff leak has only been validated as an up-front test, prior to use of steroids. The performance of the test appears to be lower after giving steroids (e.g. absent cuff leak may be more likely due to a “false-positive” such as adherent secretions around the tube).
- Studies have showed that patients treated with steroid who still lack a cuff leak may nonetheless be extubated safely, with a low risk of reintubation (Cheng 2006, Cheng 2011).
- There is no evidence that additional intervention at this point will reduce the risk of reintubation.
- Overall, the cuff leak test is a risk-stratification tool to identify patients at risk for laryngeal edema, not a pre-requisite for extubation.
If there is no cuff leak, the patient should be extubated with readiness for emergent re-intubation if necessary. If the patient was difficult to intubate initially, consider extubation with an airway exchange catheter in place to facilitate rapid reintubation if necessary (Mort 2007).
For stridor, consider treating with inhaled epinephrine, steroid, and heliox
The guidelines don't describe management of post-extubation stridor, probably because there isn't any high-quality evidence available on this topic (nor is there ever likely to be) (3). Nonetheless, this is an important issue which deserves some discussion.
Intubation vs. noninvasive support
The first consideration is whether the patient requires emergent re-intubation. Equipment and staff should immediately be prepared for intubation. Patients in extremis require immediate intubation. However, patients with mild dyspnea may be monitored carefully while attempting inhaled epinephrine and steroid therapy. Published reintubation rates among patients with post-extubation stridor average ~50%, so not all patients need intubation (Girard 2017).
Heliox may be useful to stabilize patients with mild dyspnea and stridor. This should only be done with very careful observation (e.g. in an ICU with staff immediately available for intubation). Anecdotally I've had success with this, but it remains controversial (Pluijms 2015). Careful decisions are required on a case-by-case basis (for example, based on how much distress the patient is in, how well they respond to heliox, and how difficult their airway is anticipated to be).
In adults without pre-existing upper airway issues, it is reasonable to assume that post-extubation stridor is due to laryngeal edema. If there is a suspicion of some alternative etiology, then visualization of the airway is indicated (with a bronchoscope or nasolaryngoscope). Identification of an alternative diagnosis (e.g. vocal cord paralysis, tracheal stenosis, etc.) may dramatically alter management. Airway examination is ideally done prior to reintubation, because the endotracheal tube obscures this considerably.
Nebulized epinephrine may vasoconstrict the larynx, reducing edema. As always with epinephrine, the dosing is confusing. To add an extra layer of confusion, conventionally racemic epinephrine is used for this:
- Standard/conventional epinephrine (plain old “epinephrine”) contains only L-epinephrine.
- Racemic epinephrine is a 50:50 mixture of L-epinephrine and D-epinephrine. The L-epinephrine is biologically active, whereas the D-epinephrine seems to have little or no biological activity.
Available evidence (generalizing from a Cochrane review of children with croup) suggests that there is no difference between racemic epinephrine vs. conventional epinephrine. However, the D-epinephrine in racemic epinephrine seems to be inactive, so on a mg-for-mg basis racemic epinephrine would be expected to be half as potent as conventional epinephrine.
Ultimately it doesn't matter what you use, so long as it is given promptly and in a reasonable dose. Options include the following (Pluijms 2015):
If racemic epinephrine is available and respiratory therapists are familiar with it, that's the simplest approach. Some hospitals are eliminating 1 mg/ml epinephrine (1:1000) to avoid dosing errors, so that may not always be an option (4). “Cardiac epinephrine” (0.1 mg/ml, 1:10,000) has the advantage that it's readily available in crash carts, so this always immediately available (just dump it in a nebulizer and give it, no mixing or dilution required).
There is no evidence regarding the appropriate steroid dose. A single dose of 125 mg IV methylprednisolone may be reasonable, perhaps with a dose of 40-60 mg the next day.
- The cuff leak test may be used to predict the risk of post-extubation stridor and reintubation, but its performance is limited.
- Patients at low risk for re-intubation may be extubated regardless of the results of the cuff leak tests.
- Patients at higher risk of re-intubation without a cuff leak may be managed by immediately giving a dose of steroid and delaying extubation by 4-6 hours. This balances the goals of early extubation and avoiding re-intubation.
- If post-extubation stridor does occur, it may be treated with IV steroid, inhaled epinephrine, and Heliox. Prompt management can avoid the need for re-intubation.
- An algorithm for approaching cuff leak is proposed, based on new ATS/ACCP guidelines:
- There are wide 95% confidence intervals on these likelihood ratios, so you can't take these exact numbers to the bank. This table is intended to provide a rough concept of the risk of reintubation, not exactly precise data.
- In short, what that means is that if every patient you extubate does well, you're probably being insufficiently aggressive about extubation (thereby exposing patients to risks of prolonged ventilation). Alternatively, if your reintubation rate is >> 15%, then you may be overly aggressive, extubating patients before they are ready. The concept of acceptable failure rates is explored further in this blog.
- Post-extubation stridor is a classic example of a critical-care phenomenon which is nearly impossible to study: it is rare and requires emergent therapy. It would be nearly impossible to design a prospective study of this disorder because the process of recruiting and consenting patients is too sluggish to enroll patients before critical treatment decisions are required.
- 1 mg/ml epinephrine (1:1000) is mostly intended for intramuscular injection. This form of epinephrine is highly prone to dosing errors (e.g. giving it IV rather than IM). Thus, many hospitals are eliminating this from the formulary. Some departments have replaced vials of 1 mg/ml epinephrine with epinephrine auto-injectors, to make this idiot-proof (it's impossible to inject an autoinjector into an intravenous line).