Re-analysing an 85-Patient Cardiac Anaesthesia Audit
Richard Purchon FRGS AC RODP
In 2017 I published an article examining endotracheal tube cuff pressures in patients undergoing cardiac surgery.
The original audit had a relatively simple clinical question behind it:
When an anaesthetic practitioner inflates an endotracheal tube cuff using a syringe and then judges the result by palpating the pilot balloon, how often does that actually produce an appropriate cuff pressure?
At the time, the answer was concerning.
Looking back at the original dataset nearly a decade later, however, something else becomes apparent.
The data tell a rather stronger story than I appreciated at the time.
And at the centre of it is a deceptively simple observation:
Ten millilitres of air is a volume. It is not a pressure.
Why cuff pressure matters
The cuff of an endotracheal tube performs two important functions.
It creates a seal sufficient to allow positive-pressure ventilation and helps reduce leakage of contaminated material around the tube into the lower airway.
But that seal comes at a physiological price.
The cuff sits against the tracheal mucosa. If the pressure exerted against that mucosa becomes excessive, mucosal perfusion may be compromised.
Too little pressure creates a different problem: an inadequate seal and an increased possibility of leakage or microaspiration.
The objective is therefore not simply to inflate the cuff.
It is to inflate it sufficiently while avoiding unnecessary pressure against the tracheal wall.
Traditionally, however, the method used in many anaesthetic rooms was remarkably unsophisticated.
A syringe was attached to the pilot balloon.
Several millilitres of air were injected.
The pilot balloon was squeezed between the fingers.
The practitioner decided that it “felt about right”.
The syringe was removed.
And everyone carried on.
The question is whether the human fingertip is capable of functioning as a reliable pressure-measuring device.
The original audit suggested that it is not.
The original audit
Between 2014 and 2016 I collected data from 85 patients undergoing cardiac surgery across three specialist cardiothoracic centres.
The patients were intubated as part of their normal anaesthetic management.
The endotracheal tube cuff was inflated using the anaesthetic provider’s usual technique.
Importantly, I did not tell the practitioner how much air to insert or how the cuff should be assessed.
Once they considered the cuff appropriately inflated, the pressure was measured using a cuff-pressure manometer.
The pressure was recorded and the cuff was then adjusted to an appropriate clinical pressure.
The dataset recorded:
- endotracheal tube size;
- patient height;
- volume of air inserted into the cuff; and
- measured cuff pressure.
The endotracheal tubes were predominantly size 7.0 and 8.0 high-volume, low-pressure tubes, with two size 9.0 tubes included.
The published dataset contained all 85 individual measurements, allowing the results to be re-examined rather than relying solely upon the summary contained in the original article.
Re-analysing the 85 cases
Using the pressure categories represented in the original published table, the 85 cases divide into:
12 under-inflated cuffs — 14.1%
24 cuffs within the target category — 28.2%
32 over-inflated cuffs — 37.6%
17 grossly over-inflated cuffs — 20.0%
That means only:
24 of 85 cuffs were within the intended range.
Conversely:
61 of 85 cuffs — 71.8% — were outside it.
That is a considerably more striking finding than simply saying that cuff pressures varied.
Almost three quarters of cuffs inflated according to normal clinical judgement were outside the intended range when actually measured.
An error in my original article
Re-analysis also identified an inconsistency in the published paper.
The original results section correctly listed:
- 12 under-inflated;
- 24 correctly inflated;
- 32 over-inflated; and
- 17 grossly over-inflated.
Those numbers add up perfectly to the 85 cases in the dataset.
Later in the same paragraph, however, the article states that 14 cuffs were grossly over-inflated.
That is incorrect.
The raw data support the figure of:
17 grossly over-inflated cuffs.
The same section also states that only 21% of cuffs were within the target pressure area.
Again, examination of the published dataset suggests that figure is incorrect.
Twenty-four of 85 cases represents:
28.2%.
The important point is not that the original conclusion was exaggerated.
Quite the opposite.
Even after correcting the arithmetic, the clinical finding remains substantial:
more than seven out of every ten cuffs were outside the intended pressure range.
Then comes the really interesting finding
The most commonly used inflation volume was:
10 mL
Of the 85 cuffs, 55 were inflated with 10 mL of air.
At first glance that might appear to represent reasonably consistent clinical practice.
Put approximately 10 mL into a high-volume, low-pressure cuff and assess the pilot balloon.
Except the resulting pressures were anything but consistent.
Those 55 cuffs produced measured pressures ranging from:
15 cmH₂O to greater than 120 cmH₂O.
The same nominal volume therefore produced pressures spanning the entire clinically relevant spectrum:
from under-inflation,
through an acceptable range,
through substantial over-inflation,
to pressures beyond the maximum reading documented in the audit.
Among those 55 cuffs inflated with exactly 10 mL:
- 7 were under-inflated
- 12 were within the target range
- 20 were over-inflated
- 16 were grossly over-inflated
So only:
12 of 55 — 21.8%
of the cuffs inflated with 10 mL ended up in the desired category.
More than three quarters did not.
That single observation arguably demonstrates the central clinical problem better than anything else in the dataset.
Volume does not predict pressure
It is tempting to assume that if the same model of high-volume, low-pressure cuff receives approximately the same volume of air, it should produce approximately the same pressure.
The data emphatically fail to demonstrate that.
When inflation volume was compared statistically with measured cuff pressure, the relationship was weak.
Using Spearman rank correlation:
ρ ≈ 0.15
with a probability value of approximately:
p ≈ 0.18
In straightforward clinical language:
there was no convincing relationship between how many millilitres of air were injected and the resulting cuff pressure.
That matters.
Because if volume does not reliably predict pressure, then saying:
“I always put 8 mL in.”
or:
“I normally use about 10 mL.”
is not a meaningful method of pressure control.
You have measured the volume delivered by the syringe.
You have not measured what that volume is doing to the patient’s trachea.
Why can the same volume produce such different pressures?
An endotracheal tube cuff does not exist in isolation.
Once inflated within the trachea its behaviour is influenced by several variables.
These potentially include:
- internal tracheal diameter;
- tube diameter;
- cuff design;
- precise tube position;
- patient anatomy;
- cuff folding;
- material characteristics;
- temperature;
- head and neck position;
- interaction with other devices;
- and possibly physiological changes occurring during anaesthesia and surgery.
A high-volume, low-pressure cuff is designed so that the cuff can conform to the internal surface of the trachea.
But that does not mean that a predetermined syringe volume will automatically produce a predetermined intracuff pressure.
The syringe tells you only how much gas you have inserted.
The manometer tells you what pressure resulted.
They are not interchangeable measurements.
What about palpating the pilot balloon?
This was perhaps the more important challenge posed by the original audit.
The practitioners were not blindly injecting air.
They were generally using clinical judgement.
The pilot balloon was palpated.
The cuff was assessed.
The practitioner decided when sufficient inflation had been achieved.
Yet measured pressures ranged enormously.
The obvious conclusion is that tactile estimation is a poor substitute for objective measurement.
This should not really surprise us.
If I asked a practitioner to squeeze an arterial pressure line and tell me whether the pressure was 30, 50 or 90 mmHg, we would regard the suggestion as absurd.
We measure it.
If I asked someone to squeeze a ventilator circuit and accurately estimate airway pressure with their fingers, again we would use a pressure gauge.
Yet historically we have been surprisingly comfortable squeezing an ETT pilot balloon and making essentially the same sort of judgement.
An intriguing relationship with tube size
Re-analysis produced another observation.
Smaller tube sizes tended to be associated with higher recorded cuff pressures.
The correlation was modest but statistically detectable:
Spearman ρ ≈ −0.28, p ≈ 0.010.
Patient height demonstrated a similar inverse association:
ρ ≈ −0.30, p ≈ 0.005.
Shorter patients tended, on average, to have higher measured pressures.
These results are interesting, but they need to be treated cautiously.
This was never designed as a trial investigating anatomical predictors of cuff pressure.
Tube size was not randomly allocated.
Patient sex, tracheal diameter and several other potentially important variables were not recorded.
There were also only two size 9.0 tubes.
The data therefore allow us to observe an association.
They do not allow us to claim causation.
Nevertheless, they reinforce the broader point.
The pressure produced by an inflated cuff is dependent on more than the number printed on the syringe.
The problem with “>120”
Another limitation becomes obvious when attempting modern statistical analysis of the data.
Thirteen cuff pressures were recorded as:
>120 cmH₂O
The actual pressure is therefore unknown.
One cuff might have been 122 cmH₂O.
Another could theoretically have been considerably higher.
Consequently it would be misleading to calculate and publish a precise arithmetic mean for the entire group as though those observations had exact values.
For exploratory calculation they can be treated as at least 120 cmH₂O, but doing so necessarily underestimates the true pressure of any measurement above that threshold.
The presence of so many readings beyond the recorded upper range is itself clinically significant.
A measurement problem disguised as a technique problem
Looking back, I think this is the central lesson of the audit.
The problem was not necessarily that anaesthetic practitioners were careless.
The problem was that they were being asked to determine a physical measurement without using the instrument designed to measure it.
The cuff pressure is expressed in:
cmH₂O.
A syringe measures:
mL.
A fingertip measures neither.
There may be circumstances in which experienced practitioners can approximate an acceptable seal remarkably well.
But that is not the same thing as demonstrating that the cuff lies within an appropriate pressure range.
If the value matters clinically and an inexpensive instrument exists that can measure it directly, estimation becomes increasingly difficult to justify.
What has changed since 2017?
The original recommendation was straightforward:
cuff pressure should be measured using a manometer rather than estimated from injected volume and pilot-balloon palpation.
Re-examining the data does not weaken that recommendation.
It strengthens it.
The most persuasive finding is no longer simply that some cuffs were over-inflated.
It is the extraordinary variation produced by ostensibly similar inflation practices.
Ten millilitres could produce:
15 cmH₂O.
Ten millilitres could produce:
30 cmH₂O.
Ten millilitres could produce:
70 cmH₂O.
And ten millilitres could produce:
more than 120 cmH₂O.
Same syringe.
Same nominal volume.
Completely different physiological result.
The lesson from 85 patients
Clinical practice contains many traditions that begin because they are convenient.
Some survive because they appear to work.
Some survive because experienced clinicians teach them to the next generation.
And occasionally we finally measure them.
This audit measured one.
The result was fairly unequivocal.
71.8% of cuffs were outside the intended category when objectively measured.
Among cuffs inflated with the most common volume of 10 mL, only 21.8% fell within the intended range.
And inflation volume itself showed no convincing relationship with the pressure ultimately generated.
The conclusion I reached in 2017 therefore survives re-analysis remarkably well.
I would phrase it rather more simply today:
Stop guessing.
If you want to know the volume, use a syringe.
If you want to know the pressure, use a manometer.
Because 10 mL of air is not a cuff pressure.
