Of all fields in ICU, mechanical ventilation might be the one most steeped in myth. As is common in medicine, we have stereotyped and perhaps even romanticised a necessary but invasive and often harmful practice; we have also created an arcane and confusing terminology.
Many books and articles on mechanical ventilation start with a definition of "respiratory failure" and list this as the main (or sole) indication for mechanical ventilation. Respiratory failure is then usually defined by numbers, and broken down into two forms, commonly termed "Type I" and "Type II" - the former said to be characterised by arterial hypoxia without a raised partial pressure of carbon dioxide, the latter by both low PaO2 and high PaCO2.
Think about such definitions for a moment! By looking at arterial partial pressures of gases, we are side-stepping the main issue, which is failure of the normal respiratory pump ! We are basing our definitions (and presumably our decision to ventilate) on a few observed consequences, and we are failing to take the baseline state of the individual into account.
I may seem to be arguing semantics, but .. consider a 70 year old patient with chronic obstructive airways disease who lives at an altitude of say 1600 metres above sea level. We might expect a normal seventy year old at this altitude to have a room-air PaO2 of perhaps 57 mmHg. An arterial PO2 of 47 mmHg with a PCO2 of (say) 50 mmHg tells us absolutely nothing about our patient! This person in "type II respiratory failure" might be sitting in your consulting rooms talking fairly comfortably to you, or be on the brink of intubation.
Again and again we will find that blind application of "rules" will get us into trouble with ventilatory management of patients. It is most important when assessing the illness of a patient to know something about how they normally function, and then to look at the patient. Particular attention should be paid to the respiratory rate, and the overall condition of the patient - is she distressed, sweating, poorly perfused or using accessory muscles of respiration. Such clinical observation is the correct context in which to interpret the "numbers".
We use mechanical ventilation to support patients who cannot sustain ventilation unaided. We normally do this by inserting an endotracheal tube, actively blowing air into the chest, and then allowing the air to passively leave. In so doing, we create a multitude of problems, including:
We have invented a vast variety of different ways of mechanically ventilating patients, but in adults there is no clear evidence that any one method is superior to the others. Clearly some are more comfortable to the patient than others, but even here, there is substantial variation. Dogmatic assertions that things must be done in a certain ritualised fashion are more commonly based on arrogance than evidence. Nevertheless, over the past ten or so years, we have perhaps progressed a little in defining approaches to ventilation that cause less harm than we in our ignorance used to cause! This web-page looks at some of the speculation and facts surrounding these 'new' approaches.
Before we being, we might briefly describe a framework that is useful in describing ventilation. This isn't as easy as it sounds, as there is little consensus on how to classify the different modes of ventilation. We'll try not to get too bogged down. Then we will go through various ventilatory strategies that have shown promise in the past few years. These include:
Ventilator terminology is confusing, especially as some authors have used the same words - "control", "assisted", "spontaneous" and "mandatory" - for different things! We would like to draw your attention to the way Kapadia has simplified ventilator terminology [Postgrad Med J 1998 74 330-5]. He uses three basic terms:
This 'TLC' terminology has the merit of being simple and fairly comprehensive. Triggering is straightforward - either the patient initiates a breath, or the machine automatically does so at a given rate. (Different signals from the patient may be used to trigger a breath - older ventilators relied on changes in pressure, but more recently the sneaky approach of detecting changes in flow has been used to decrease the effort that a patient has to employ to trigger a breath
Cycling is complex, as there may be a pause between the cessation of inspiration, and the onset of expiration (as the expiratory valve opens). Without a pause, the signal that ends inspiration is immediately used to trigger expiration; otherwise, the machine inserts a delay between closure of the inspiratory valve, and opening of the expiratory valve. Cycling may be triggered by:
Kapadia conveniently describes the commonly-used modes of ventilation in his Table 1, which we have abbreviated as follows:
|Common modes of ventilation - TLC classification|
|Continuous Mechanical Ventilation Assist (CMVa)|
= Assist Control (A/C)
= Volume Control Assist (VCa)
|Ventilator or Patient||Flow||Volume (Time controls pause)|
|Pressure Control Ventilation (PCV)||Ventilator or Patient||Pressure||Time (Time also controls pause)|
Synchronised Intermittent Mandatory Ventilation (SIMV)
|Ventilator or Patient||Flow (mandatory breath)||Volume (mandatory breath)|
|pressure-limited SIMV||Ventilator or Patient||Pressure (mandatory breath)||Time (mandatory breath)|
|Pressure Support (PS)||Patient||Pressure||Flow|
|CPAP + PS||Patient||Pressure||Flow|
|SIMV + PS||A combination of synchronised intermittent mandatory ventilation (with the appropriate characteristics of the mandatory breaths) and pressure support (with its characteristics). Note that either type of SIMV mentioned above may be used.|
|Note that where CPAP is combined with ventilator triggered modes, confusing terminology kicks in again - CPAP is then called "PEEP" (Positive End-Expiratory Pressure).|
A/C is a mode that is really only used in theatre, and is an anachronism in the ICU. A fixed number of breaths is delivered, but the patient may also trigger delivery of a full breath. PCV used to be seen as a "last resort" in patients with very stiff lungs, running high pressures on SIMV". Many intensivists now view it as a good primary mode of ventilation in patients with poor lung compliance (for example, ARDS), especially when combined with longer inspiratory times (inverse ratio PCV, otherwise known as "PCIRV"). In SIMV , breaths are delivered by the machine at a preset rate, and the patient can initiate an (unassisted) breath in between the machine breaths. If the patient starts breathing just before a machine breath was due, then only is a full machine breath delivered. Note that with pure PS or pure CPAP no machine breaths are delivered - the patient initiates all breaths. On many machines (especially older ones) the work of breathing in CPAP mode is so substantial that this mode should not be used!
In the following sections, we look at new approaches to ventilation
(including new modes). We will refer to their TLC characteristics as
we examine these modes.
1. Altered pressure profiles and related strategies
Here we will examine two strategies, the first being the use of so-called
"BiPAP", and the second the use of feedback to dynamically alter
There are at least five circumstances where the dreaded term "BiPAP" has been used:
Airway Pressure Release Ventilation (APRV) appears to have had a mixed reception. Some experimental studies show little merit (APRV proved inferior to CPAP in maintaining oxygenation and lowering shunt in an oleic acid model of ALI in pigs [Anesth Analg 2001 Apr;92(4):950-8]), while others disagree, showing improved VQ mismatch with unrestricted spontaneous breathing during APRV compared with pressure support [ Am J Respir Crit Care Med 1999 Apr;159(4 Pt 1):1241-8]. Isolated case reports claim benefit in a variety of respiratory disorders.
The old-fashioned term "servo" (which just means feedback) is commonly used in ventilators that use feedback to alter settings according to how the patient's lungs respond. In each case, we must define the parameter that is monitored, and the feedback response (alteration) that occurs:
During spontaneous breathing, the resistance of an endotracheal tube
may contribute substantially to the work of breathing, especially at
high flow rates, and with smaller tubes. With 'ATC', inspiratory pressure
is increased to compensate for the resistance of the tube, and this
compensation is varied with flow, providing full compensation. The
work that the patient has to exert in overcoming the resistance of the
tube effectively 'disappears'. It has even been claimed that because of
this lack of tube resistance, ATC "predicts the patient response after
extubation"! (Evita ventilators).
2. Fast Rates
Several modes of high-frequency ventilation have been proposed, including
high-frequency oscillatory ventilation (HFOV), and high-frequency jet
ventilation (HFJV). HFJV has been mainly used during short procedures
such as bronchoscopies. It is not without complications, notably
pressure "stacking", hypercarbia, and airway injury owing to difficulty
with gas humidification. High-frequency ventilation can be considered
ventilator-triggered, flow (HFJV) or pressure (HFOV) limited, and
time-cycled. Some commercial machines allow rates of up to about forty
breaths per second (2400/min), although substantially slower rates
are usually employed, in the range of 2-15 Hz.
HFOV is interesting. It was first thought of as a ventilatory modality when someone noticed that dogs, when they pant , take breaths that are smaller than their dead space. (Dead space in all animals, even giraffes, is generally constant at about 1/3 of resting tidal volume). How, the researcher wondered, do panting dogs maintain oxygenation? To this day, we're still not sure of the answer - at least five different credible explanations have been proposed. Simplistically, the high frequency of panting (and HFOV) increases turbulence and thus mixing and diffusion of oxygen.
HFOV has been extensively used in neonates, and has almost become a 'standard of care' in some centres. There is however controversy about when it should be used, and even whether it is beneficial. Recent meta-analyses question its benefit in neonates, especially now that surfactant can be delivered into the lungs of infants with surfactant deficiency. There is some evidence to suggest that HFOV may be associated with an increased incidence of severe intraventricular haemorrhage, and periventricular leukomalacia.
The largest HFOV trial in neonates (the "HIFI" trial) had negative
results, but the methodology employed has been questioned. Clearly, HFOV
is tricky. There is experimental evidence to suggest that HFOV at low
mean airway pressures may be extremely harmful to the lung, especially
if lung compliance is poor, causing extensive barotrauma. "High-volume"
strategies are now advocated (i.e. using high mean airway pressures).
Inappropriate HFOV strategies probably account for the poor results
initially encountered with HFOV in adults. HFOV may well still have
a place in ventilation of adults with severe ARDS, but no decent-sized
studies have been performed. We wait!
Many have written about the value of positive end-expiratory pressue
(PEEP/CPAP) in critically ill patients, especially in ARDS. There is
fairly strong evidence that PEEP is of great value in the management of
ARDS, but how to obtain "best" PEEP is very controversial. Everone has
his/her own approach! Some have empirically taken +2cm H 2 O above the
"lower inflection point" (LIP) of a quasi-static pressure-volume curve.
This is performed by paralysing the patient and then slowly expanding
the lungs using a "super syringe", all the time monitoring airway pressure
and volume injected. There are numerous problems with this approach
(apart from the time taken and the requirement of total patient paralysis).
Among the problems are:
Rigorous studies looking at outcome as related to level of PEEP are
lacking. Most approaches are anecdotal. Our particular anecdotal approach
is to use at least 10cm H 2 O PEEP unless there is a good reason to not
do so! Patients who become cardiovascularly compromised at such levels
are usually grossly behind on fluids. Exceptions to this approach should
include asthmatics (where one should generally shy away from PEEP), and
patients with chronic obstructive airways disease. Note that in asthmatics
with residual auto-PEEP despite slow rates (say 8 breaths per minute),
judicious application of a small amount of PEEP to balance the auto-PEEP
may in fact be beneficial! Take care with PEEP (although it's probably
a good thing overall) - recent work suggests that although it
works in experimental models of ARDS, its success is at the expense
of increased parenchymal stress [ J Appl Physiol 2001 May;90(5):1744-53].
4. "Non-invasive Ventilation" and Negative Pressures!
Non-invasive positive-pressure ventilation (NIPPV) has been around
for ages, but earlier masks were hellishly uncomfortable, and patient
selection was often poor.
Kacmarek has reviewed
the "level one" evidence that NIPPV works. NIPPV appears to be
particularly advantageous (with careful selection) in patients with
chronic obstructive pulmonary disease. For a meta-analysis, see
[Crit Care Med 1997 25 1685-92]. It has also been used in cystic
fibrosis, acute asthma, and patients awaiting lung transplantation.
NIPPV may be life-preserving in immunocompromised patients with
Pneumocystis carinii or other pneumonias
[Chest 1996 109 179-93;
Intensive Care Med 1998 24 1283-8;
JAMA 2000 283 235-41].
The major advantage of NIPPV is probably the lack of an endotracheal tube -
normal defence mechanisms that prevent entry of bacteria into the lung
[Intensive Care Med 1999 25 567-73;
Ann Intern Med 1998 128 721-8;
N Engl J Med 1998 339 429-35]
There is now compelling evidence that early extubation to NIPPV lowers duration
of mechanical ventilation, shortens ICU stay, lowers the incidence
of nosocomial pneumonia and improves sixty-day survival
[Ann Intern Med 1998 128 721-8].
There are still numerous 'tricks' to establishing successful NIPPV. The patient must be co-operative, and gently informed that you have a mask that will help their breathing. NIPPV must be mechanically feasible. Don't show the patient the head-gear first off - start with a low level of CPAP/PEEP (± PS), and allow them to hold the soft silicone mask gently to their own face. When you have their confidence, then you can adjust settings to an optimal level, and finally strap the mask on. It's often initially quite tricky (or impossible) to get a distressed patient to breathe through a nasal mask - the drive to breathe through the mouth may be overwhelming! NIPPV may make large demands on nursing staff, although this has generally not been the case. Successful outcomes tend to correlate with the presence of teeth, younger age, lower APACHE/SAPS scores, and less acidosis. Guidelines (which seem to be a bit of a bland thumb-suck) are:
|Consensus Guidelines: NIV for COPD + Acute Respiratory Failure|
|Step 1: Identify need for ventilatory assistance
|Step 2: Exclude those at high risk
|Modified after [Respir Care 1997 42 364-9]|
The first ventilators were tank ventilators such as the "iron lung"
that relied on extracorporeal application of negative pressure,
with a rubber seal around the patient's neck. These fell into disfavour
for a variety of reasons including upper airway obstruction (obstructive
sleep apnoea, even in "normal patients"), claustrophobia, discomfort,
lack of triggering in response to patient attempts to breathe, leaks,
and sheer bulk (the "iron lung" weighed a ton, even modern cuirasses
are bulky). Newer developments in this field have yet to be extensively
tested. A recent, thorough review of all aspects of non-invasive ventilation
[Am J Respir Crit Care Med 2001 163 540-77]
includes a good history of development of tank and cuirass ventilators.
5. Liquid Ventilation
There is a vast amount of experimental literature on liquid ventilation.
Most commonly, this is a combination of airway instillation of
oxygen-carrying fluorocarbons such as perflubron, with conventional
ventilation (partial liquid ventilation, PLV).
Unfortunately, clinical studies are few and far between.
There is a lot we still need to learn about PLV - for example, distribution
of perflubron within the lung is far from homogeneous, tending to gravitate
towards dependent areas. Initial improvements in oxygenation need not
necessarily be maintained, and falsely high tidal volumes may occur
(fixed orifice measurements are up by 7-16%, hot wire 35-41%) [Physiol Meas 2000 Aug;21(3):N23-30]
Claimed merits of PLV include improved oxygenation, wash-out of exudates
and infectious material, decreased bacterial adhesion
[Crit Care Med 1999 27 2741-7],
and even (possibly) reduced lung injury [Crit Care Med 1999 27 2500-7].
6. Altered Volumes - "Protective Strategies"
The ARDSNET study is probably the most important paper on ventilation from
the last decade. This well-designed study demonstrates unequivocally
that lower tidal volumes (6ml/kg ideal weight) are associated with
improved outcomes, including 30-day mortality, when compared with
more "conventional" volumes (12ml/kg). Even more interesting is the
observation that this improvement was in the face of slightly lower arterial
partial pressures of oxygen. Also of note is that in this study,
pH was corrected with intravenous sodium bicarbonate, to remove pH
as an interfering variable. Another interesting question is that of
respiratory rate - some have argued by unconvincing analogy that
respiratory rates should be kept low in ARDS, but the ARDSNET group
that did better had average rates of 29/min (as opposed to half this
rate in controls).
The ARDSNET study is the most important of several studies which have looked at the question of tidal volumes in ARDS, some of the others agreeing with its findings, and other coming up 'negative' for reasons capably discussed in the paper itself [New Engl J Med 2000 May 342 (18) 1301-9].
By the way, note the introduction of the peculiar term "volutrauma".
This is a silly and quite unnecessary neologism, as pressure and volume
are intimately related. The term seems to arise from a failure to
understand that there is an inconstant relationship between transalveolar
pressure and transthoracic pressure. Whatever the terminology used,
it is now quite clear that alveolar overdistension results in disruption
of the alveolar capillary membrane, and florid pulmonary inflammation.
7. Altered Goals - O2, CO2 and pH
Only recently have we really begun to appreciate that "normality" is a far
from reasonable goal in the critically ill. Everything in ICU is a trade-off,
and previously we were probably far too enthusiastic in our attempts to
reach a "normal" saturation of say 93(+)%, a "normal" PCO2 of say 40 mmHg,
and a "normal" pH of 7.36 to 7.44.
There is emerging evidence that we may be wrong on all three counts:
Note that permissive hypercapnia induces dramatic changes in the
[Am J Respir Crit Care Med 1997 Nov;156(5):1458-66]
and may also tend to worsen PaO2, as shown
in a multiple inert gas study on a small number of patients
[Am J Respir Crit Care Med 2000 Jul;162(1):209-15].
Correcting the pH does alleviate the haemodynamic effects of permissive
hypercapnia [Crit Care Med 1996 May;24(5):827-34], at least in sheep.
Some have advocated the use of THAM as an alternative to bicarbonate
administration [Am J Respir Crit Care Med 2000 Apr;161(4 Pt 1):1149-53].
8. Prone Positioning
In the absence of large studies on prone positioning, one has to be content
with empiric generalisations. Prone positioning does appear to benefit
a proportion of patients with severe ARDS.
[Intensive Care Med 1997 23 1219-24;
Anesth Analg 1995 80 955-60;
Anesthesiology 1991 74 15-23;
Am Rev Respir Dis 1987 135 628-33]
The reason for this is not clear,
but may be to do with the heterogeneous nature of the lung injury -
"ARDS" seems to preferentially affect the dependent regions of the lungs,
so a patient lying on their back may benefit from improved V/Q ratios when
flipped over. The benefit of prone positioning may take several hours
to become apparent. Nursing is not generally a problem once the nurses
have become accustomed to this approach, (but one must take care not to
allow undue pressure on the eyes - this should be common sense). It
makes sense to try this maneuver in ARDS patients with refractory arterial
hypoxia, before indulging in more aggressive heroics. Prone positioning
is also an important component of recruitment (see below ).
9. Low Dead Space and Tracheal Gas Insufflation
One problem with critically ill patients, especially those with
severely diseased lungs, is high VD/VT ratios (large dead space:
tidal volume). This may be a substantial problem as we decrease tidal
volume in an attempt to limit barotrauma, although most patients
tolerate hypercapnia fairly well. Exceptions would be patients with
severe associated metabolic acidosis, or those with raised
intracranial pressure. Possible solutions are
to use specially constructed endotracheal tubes with an internal
second lumen to minimise dead space, or even to insufflate gas via
a small cannula. Although attractive in concept, there seem to be
significant problems, including:
See [Anesthesiology 1997 87 6-17;
Am Rev Respir Dis 1993 148 345-51]
10. Recruitment Maneuvers
A variety of approaches have been used to "recruit" collapsed air spaces
in the lung. The general consensus seems to be that substantial pressures
(40+cmH 2 O) are needed to open up such air spaces - simply cranking up
the PEEP a few centimetres is not usually adequate, although such
"mini-recruitment maneuvers" may cause mild improvements in
oxygenation [Intensive Care Med 2000 May;26(5):501-7].
We have elsewhere discussed one approach to recruitment that seems particularly
effective. All such approaches should at present be regarded as "experimental",
although the results of a successful recruitment maneuver are so dramatic
and satisfying that we cannot countenance not trying this maneuver in
severely hypoxaemic patients. We can only hope that formal studies of
recruitment don't hash things up (and give false negative results)
by, for example, failing to turn the patients prone before recruitment
11. Paralysis, Sedation, and the Myth of Weaning
As always, one's goal in ventilating a patient should be to
minimise intervention and yet support the patient adequately. An awake,
comfortable, co-operative patient is infinitely preferable to a heavily
sedated patient, which in turn is far, far superior to having to
paralyse a heavily sedated patient. We have only recently begun to
realise the carnage that we wreak when we paralyse a patient for a
prolonged period (over 24 hours) - so-called "critical illness polyneuropathy"
(and/or myopathy) is a dramatic and devastating consequence of aggressive
neuromuscular blockade, especially if level of paralysis is not monitored using
a nerve stimulator, and/or steroids are administered concomitantly.
The term "weaning from the ventilator" is unfortunate. It conjures up the image of ventilation as something akin to mother's milk, from which we reluctantly 'wean' the patient. This picture may be far from the truth.
It would appear logical to remove the invasive and noxious intervention of ventilation as soon as the patient is ready to have this removed. Here is the catch - our "weaning strategies" are often merely maneuvers that we indulge in to reassure ourselves that the patient is ready to have the ventilator, and finally, the endotracheal tube, removed! (Some have called this "liberating the patient from the ventilator", which I find far more honest)!
Clearly however, as a patient improves, modification of our ventilatory
strategy may make the patient more comfortable, and this is
desirable. There are no cast-iron rules. For example, in some
intensive care units, pressure control ventilation is still regarded
as an approach of last resort, and heavy sedation and paralysis are
regarded as necessary concomitants! As the patient improves, he/she
is then put "back on SIMV+PS", and the SIMV rate is then
progressively "weaned". There is no justification for this approach,
(although it often works as well as any other). You will find, with
careful experimentation, that some patients on pure pressure support
will be more comfortable on pressure control ventilation,
as with the latter one can set an I:E ratio (cycling is time-dependent
and not flow-dependent). Others will be very intolerant of pressure
control. Often simple observation of the patient (looking carefully
at the pattern of ventilation) will allow one to adjust ventilation
appropriately. Newer modes of ventilation such as APRV, volume support
and PRVC may be extremely useful in improving patient comfort.
An Approach - Ten Guidelines
What can we conclude from the above? For patients with severe lung
disease, there is no one approach - diseases differ, and so must
managment. The following seem to us to be reasonable guidelines:
|Date of First Publication: 2001/5/15||Date of Last Update: 2006/10/24||Web page author: Click here|