Post-extubation stridor

[toc]

Post-extubation stridor is defined as the presence of an inspiratory noise following extubation. Colloquially, it is believed to be the consequence of some sort of narrowing of the airway, resulting in an increased effort of breathing. The usual site of narrowing is the larynx, and oedema is blamed as the underlying pathology. Methods of managing this complication are mainly anti-inflammatory (eg. dexamethasone and nebulised adrenaline) or mechanical (eg. NIV and heliox).

Question 27 from the second paper of 2008 and Question 7.2 from the first paper of 2010 have raised this issue in the written papers. The former was straightforward (risk factors and management) whereas the latter offered stridor in the form of a young man with severe airway injuries.

The most relevant literature sources for this summary were the 2009 review article by Wittekamp et al as well as its update 2015 by Pluijms et al.  For exam purposes, most people should stop reading there. An even better move would be to stop reading here and absorb the succinct LITFL summary of this topic.

Definition of post-extubation stridor

There does not seem to be any widely accepted definition, but most people agree that for the term to be used appropriately there should be a characteristic piteous choking sound made by the increasingly hypoxic patient, which stimulates the surrounding medical staff to prepare for reintubation.

Historically, learned gentlemen of science have defined post-extubation stridor in a number of different (occasionally surprisingly poetic) ways. The following list is selected randomly from the literature and  Wittekamp et al (2009):

• "High-pitched inspiratory wheeze within 24 hours of extubation with respiratory rate >30" - Maury et al, 2004
• "Inspiratory grunting, whistling or wheezing requiring medical intervention within 24 hours after extubation" - Kriner et al, 2005
• "Crowing sound on inspiration" - Ho et al, 1996
• "Near total occlusion of the airway as seen on video bronchoscopy" - Chung et al, 2006
• "A musical, continuous sound" on inspiration, which is "more intense over the neck" - Baughman et al, 1989

It seems the authors disagree as to what the sound should sound like, whether or not a sound is required (total occlusion would be silent), how the obstruction is confirmed (bronchoscopy?) or what degree of respiratory distress is viewed as clinically significant. The only thing they all seem to agree on is that the problem is inspiratory, as opposed to wheeze which is an expiratory phenomenon.

Risk factors for post-extubation stridor

After four days or so of intubation, laryngeal oedema develops in the majority of ICU patients (94% according to Colice et al, 1989). The incidence of stridor after extubation is of course much lower than that. According to Wittekamp et al (2009), it is clinically significant in 30%, and leads to reintubation in 4% of patients.

The major risk factors for post-extubation stridor listed below have been pillaged from Table 2, Pluijms et al (2015). They are as follows:

• Prolonged ventilation
• Female gender
• Under-sedation (i.e. insufficiently deep; too awake)
• Difficult intubation (multiple attempts)
• Self-extubation
• High BMI (over 26.5)
• Ratio of tube size to laryngeal size in excess of 45%
• High cuff pressure
• High SAPS II score (i.e. severe illness)
• Medical patient (i.e. it was not an elective perioperative intubation)

LITFL also list the following risk factors:

• prolonged intubation attempt (>10min)
• oroendotracheal intubation
• larger tubes
• short neck
• trauma patients
• known airway pathology (tracheal stenosis, tracheomalacia)
• children
• small height:internal diameter ETT ratio
• agitation while intubated
• recurrent intubations

Finally, the college answer to Question 27 from the second paper of 2008 lists several more, which have not been mentioned in either the old (2009) ore the more recent (2015) systematic reviews.

• Trauma, surgery or infection of upper airways
• Older age group
• Elevated APACHE

It is hard to tell where these came from. Wherever it was, all these associations can hardly be described as scientific. For instance, let us take duration of mechanical ventilation. Some authors  say 36 hours is the risk zone, some say over 6 days; the college answer for Question 27 from the second paper of 2008 says 5 days, and so is probably the definitive number for future answers. However among the authors quoted by  Pluijms et al (2015) about half failed to show any relationship between duration of intubation and post-extubation stridor. Does a relationship exist, then? It is probably more complex, and depends on what happens during the time spent ventilated. Clearly, if you lay there obediently for the whole five days with the head of your bed up at 45°, your extubation will be somewhat different to the guy who is severely agitated or the one who has been prone-ventilated for sixteen hours a day.

Anyway. For the purpose of simplified revision, risk factors and causes for post-extubation stridor can be separated into the following groups:

Risk Factors and Causes of Post-Extubation Stridor

Tube factors	Large ETT for airway (eg. dual-lumen) Ratio of tube size to laryngeal size in excess of 45% Prolonged intubation (>5 days?) High cuff pressure
Airway factors	Small airway Female gender (small airway) Small height:internal diameter ETT ratio Short neck Pre-existing airway pathology, eg. tracheomalacia
Causes of airway trauma	Airway injury or surgery (eg. dental abscess) Difficult intubation (multiple attempts) Reintubation Self-extubation Agitation, under-sedation
Patient factors	Old age Overweight (BMI >26.5) Under-sedated
Disease factors	High SAPS-II High APACHE score Medical patient

Predicting post-extubation stridor: the cuff leak test

The cuff leak test is a crude experiment, performed upon the subglottic tissues. The cuff occupies a space somewhere in the mid trachea. If the cuff is inflated, gas from the lower airway should not leak up into the upper airway, and the seal which is formed in this manner more or less defines the whole purpose of the cuff.

So, if you let the cuff down, a few things will happen:

• A satisfying gurgling sound will be audible, coming from the patient's mouth
• The ventilator will alarm, and issue a calculated leak volume which represents the difference between inspired and expired circuit gas volumes
• Some crusty filth collected in the subglottic space above the cuff will drop into the trachea
• The patient will cough violently, as the ETT comes into closer contact with the tracheal wall

How to perform a formal cuff leak test

The cuff leak test has been formalised into something of a ritual, so that all researchers might follow roughly the same standard. The effect of this is such that  LITFL and Pluijms et al (2015) offer a description of essentially the same procedure:

• Set the ventilator to assist-control (i.e. a volume-controlled mandatory mode).
• Suction oral and tracheal secretions (see the crusty filth comments above).
• Record displayed inspiratory and expiratory tidal volumes.
• Deflate the cuff by 10ml
• Observe the leak over the next six breaths (apparently the expiratory tidal volume will reach a plateau value after a few cycles, and you should not simply accept the first leak volume).
• Record those six leak volumes
• Average the three lowest values
• The cuff leak is the difference between the inspiratory tidal volume and the
• Directly record the expiratory tidal volume over the next six breathing cycles as the expiratory tidal volume will reach a plateau value after a few cycles.

This formal description of the cuff leak test comes from a 1996 article by Rachel Miller and Randolph Cole, whose attempt to associate leak volume with risk of stridor was probably the first of its kind. Previous studies seem to have relied on leaning in real close and listening for an audible gurgle, sometimes with the ETT occluded so the patient is forced to draw breaths around it (Fisher and Raper, 1992). Others have actually expected their patients to vocalise audibly  around the deflated cuff (eg. the croupy children from Adderley and Mullins, 1987).

The meaning of the cuff leak volume

Having dutifully recorded your six expiratory volumes and calculated the cuff leak, one is left with a volume of some sort. In their group of 100 intubated patients. Miller and Cole found that the volume with the best predictive value for post-extubation stridor was 110 mls. Of the patients with a leak of less than 110ml, 67% developed stridor after extubation. Of the patients with more than 110ml, practically nobody developed stridor. Critically, among Miller and Cole's group only 6 patients out of the 100 ended up developing stridor, which was not a very large population. The mean leak was actually around 350ml.

Pluijms et al (2015) offer a table (Table 4) which lists the subsequent research in this area. Among the ten studies listed, the cut-off values for a cuff leak have ranged from 88 ml to 140 ml. One article (Maury et al, 2004) even found that 0 ml was a sufficiently discriminating cut-off, i.e. that any leak whatsoever predicted a successful extubation in the majority of cases. LITFL offer 10% of inspired tidal volume as the leak cut-off.

All authors seem to agree: the absence of a leak does not preclude the possibility of a successful extubation. For instance, in the abovementioned series by Maury et al 20% of the successfully extubated patients (n=111) had no cuff leak. In the presence of a cuff leak, one may be reasonably confident that post-extubation stridor will not occur.

Management of post-extubation stridor

Corticosteroids

The use of corticosteroids as a preventative pre-extubation measure is a well known trick. The college answer for  Question 27 from the second paper of 2008 mentions "a recent Lancet paper", by which I expect they mean  François et al (2007). That was the ARCO study, a large (n=761) randomised trial which compared three doses of 20mg methylprednisolone to placebo. The steroids were given every four hours for 12 hours prior to extubation; the patients had to be ventilated for over 36 hours to be eligible. The use of this methylprednisolone dose had markedly reduced the incidence of laryngeal oedema (3% vs 22%) and halved the reintubation rate (4% vs 8%).

Locally, dexamethasone is use in preference to methylprednisolone. The dose is supposed to be 0.15mg/kg, which would be about 12mg over a 24 hour period for a 70kg person -  but one frequently sees doses at least twice as high. For instance, the trial by Lee et al (2007) used dexamethasone doses of 5mg every six hours for 24 hours prior to extubation (a total of 20mg, closer to 0.30mg/kg). It is hard to argue with success: in this group of patients with inadequate (<110ml) cuff leak the rate of post-extubation stridor decreased from 27% to 10% with this sort of dexamethasone overdose.

That said, steroids have not been found to be uniformly beneficial, probably because the trials of steroids have not been uniform in their choice of drugs, dose regimens, or patient selection. Some trials enrol all patients, others only select those with low cuff leak volumes. Some give methylprednisolone in 20 or 40mg doses, others use hydrocortisone or dexamethasone. Pluijms et al (2015) offer a summary of what is currently known:

• Best to give steroids 12-24 hours prior to the extubation attempt (trials of single-dose regimens given one hour prior to extubation did not show any benefit)
• 20mg Methylprednisolone as 3 4-hourly doses is an appropriate choice, following François et al (2007)
• The effect seems to develop about 7 hours after the steroids are started
• Multiple doses seem to be better than single doses
• The benefit is greatest in patients at greatest risk of laryngeal oedema, specifically those who have low cuff leak volumes (i.e. you should not be giving prophylactic steroids to all pre-extubation patients)
• One should keep in mind that many patients with low cuff leak volumes can be extubated safely without steroids, and so using the cuff leak alone to target your steroid therapy will result in an over-use of steroids.
• As to which steroid, "We suggest a dose of 0.5 mg/kg prednisolone intravenously per day" say the authors.

Nebulised adrenaline

The use of a potent local vasoconstrictor to decrease laryngeal blood flow and therefore oedema seems like a sensible answer to post-extubation stridor. Provided the patient is not in a profoundly cyanosed coma, this is viewed as a good way to kill time while setting up for a re-intubation. Over the five minutes it takes to administer, either the patient gets better, or you have your drugs drawn up and equipment readied.

This practice of nebulising adrenaline to manage stridor is actually borrowed from paediatric practice, where it enjoys popularity as a treatment for viral croup. da Silva et al (2012) investigated the utility of this technique in extubated children via a randomised controlled trial. Extending the connection with croup, they used a well-known croup severity score (the Westley Croup Score) which counts stridor as only one domain among several markers of respiratory distress. Somehow, in this group there was a complete lack of benefit. Only undesirable effects seemed to increase in response to escalating adrenaline doses.

In adults, there is even less evidence. Pluijms et al resort to quoting a case series of 4 patients from 1995 in support of this practice. Adrenaline nebs also have cautious support from the Difficult Airway Society ("may reduce airway oedema", they say).  Nobody knows what dose to use or how long to continue. LITFL recommend 0.5ml/kg of 1:1000, up to a maximum  of 5ml. The case series mentioned above used 1mg dispersed in 5ml of normal saline.

Non-invasive ventilation

CPAP and NIV in general are mentioned as rescue strategies for post-extubation stridor. The college have included it as a part of their model answers. Again, this seems to be an extrapolation of a practice from other areas. Consider:

• High PEEP helps people overcome laryngospasm following anaesthesia, and laryngospasm is functionally a lot like laryngeal oedema.
• PEEP helps reduce the work of breathing in wheezy asthma, so t should help stridor also, because stridor is ...a kind of wheeze... right?

The theoretical benefit of PEEP is the constant pressure on the airway, and therefore a theoretical "splinting" effect on the oedematous larynx. Does this actually happen? Nobody really knows. Evidence regarding this practice is based on weak and contradictory studies. In a randomised controlled trial by Esteban et al (2004) noninvasive ventilation did not reduce mortality or the need for reintubation in patients who developed post-extubation respiratory failure. In fact, the NIV group had slightly higher mortality.  Even though they did not address stridor in any sense (it was not among the data they collected), Pluijms et al quote this trial as a part of a suggestion that NIV might inappropriately delay the inevitable reintubation. In contrast, a meta-analysis from 2013 (Krishna et al) found some mortality benefit from using "prophylactic" NIV on all patients at the time of extubation.  In short, the role of NIV in stridor is unclear. It may be a time-wasting strategy, or nothing more than a convenient way of administering 100% oxygen prior to reintubation.

Helium-oxygen gas mixture

When one's airway is too narrowed to function normally, the mind naturally turns to ways of cheating against airflow resistance. One such way is to artificially reduce the density and viscosity of the gas mixture. The ideal mixture is 30% oxygen and 70% helium, which makes it difficult to use in severely hypoxic patients. 

This is not a new thing. It seems to be well known in the paediatric circles. In the review article by Gupta et al (2005), it is discussed with reference to croup as well as other forms of stridor. In the post-extubation setting, it is known from case reports. If an experienced intensivist reaches deep, they can usually recall such cases ("remember back in '09 when we..." etc). Probably the best collection of such anecdotes is an ancient article by Skrinskas et al (1983), reporting the use of heliox in ten patients between the years 1976 and 1980. A few of them had post-extubation stridor, whereas others had airways obstructed by some sort of tumour.

Modern evidence for this practice is lacking. There are only a few situations where one might take this option instead of the others mentioned here. The author will now attempt to stretch the imagination of the readers by offering a series of potential scenarios where heliox may be appropriate:

• The patient has clinically significant stridor with respiratory distress but NOT hypoxia.
• The patient is not a candidate for NIV (eg. facial fractures, oesophageal surgery, etc).
• The patient is not a candidate for reintubation (eg. there is a limitation of therapy order).
• There is contraindication to nebulised adrenaline (eg. HOCM, or severe untreated ischaemic coronary artery disease)
• The patient has already had corticosteroids, with zero effect

Just put the tube back in

Why not just reintubate them? Well. The major argument for the use of CPAP, heliox and all the other rescue therapies is that the oedematous larynx is not going to get any less oedematous by your action of repeatedly cramming tubes though it.

The disadvantages of reintubation are mainly related to increased trauma. The extubation attempt may also have left the patient with substantially worse lung function (eg. due to collapse) With another few days spent on the ventilator, the oedema may get worse, essentially committing the patient to a tracheostomy. Ultimately, most people would wait around, trying a few of the rescue therapies first. Exactly how long you wait depends on the response to these therapies as well as on the rate at which they develop cyanosis and coma. Most people tend to wait calmly, knowing that even severely obstructed breathing is reasonably safe in a well-monitored environment.