The question of how to select the ideal PEEP for a patient, or how one would set up the ventilator for an ARDS patient, has not come up in the papers up until Question 29 from the second paper of 2016. It seems like something ICU trainees should probably be well-acquainted with, so that - when posed with the question "how do you decide what PEEP to set" - they do not dissolve into a puddle of warm gibberish.
If one were pressed for time, the best article to read about this topic would probably be Gattinoni et al (2015), as it deals with this problem in a structured and straightforward manner. It almost resembles a college model answer in the way it is laid out. This article was the main resource for the summary below. Additional short notes can be found at the LITFL CCC archive.
In brief, the possible approaches are:
Good question. What is the point of this whole chapter? Well. In short, the aim is to achieve open-lung ventilation. However, there is no such thing as the "optimal" PEEP, and therefore there is no optimal method for determining this mythical PEEP value. But, at some point you need to decide on your ventilator settings.
What makes the PEEP "optimal"? The setting is said to be optimal when:
Some of the methods of PEEP selection aim for oxygenation as their primary goal, whereas others are more about lung-protectiv ventilation. Neither approach is perfect, and each has its merits.
This is basically the lazy man's approach. You convince yourself that there is no good data to favour between one method of determining PEEP as compared to another, and you come to the conclusion that it's all bullshit. The empirical evidence seems to demonstrate that higher PEEP is associated with improved survival, so why not set a nice high PEEP and forget about the academic gibberish.
How high, you ask? There are three trials to examine: ALVEOLI, LOVS and EXPRESS. The findings have been complied into a nice meta-analysis. The high-PEEP groups averaged around 12-15 cmH2O; the low-PEEP groups had about 8-9 cmH2O. Though overall there was no in-hospital mortality benefit, the authors were forced to conclude that there seems to be a small (5%) mortality benefit for the most severe groups, i.e. those with a PaO2/FiO2 ratio less than 200.
In short, the more severe your ARDS, the higher a PEEP you ought to use, up to a possible limit of around 15. Though the ARDSNet protocol goes all the way up to 24, one might consider stopping short of this value (indeed many ventilators stop at 20 cmH2O). Gattinoni et al (2015) suggests setting 15-20 cmH2O for severe ARDS patients "may be a reasonable approach".
This resembles the lazy approach. One does not select an arbitrary PEEP: one instead looks up a table of seemingly arbitrary PEEP values on the ARDSNet protocol card:
In this protocol, worsening hypoxia is matched by an escalation of PEEP, from a PEEP of 5 cmH2O at 30% FiO2 to a PEEP of 22-24 at 100% FiO2. These values are the same as those used in the ARMA trial, and this forms the basis for their popularity. The landmark trial from 2000 demonstrated a substantial mortality improvement in the group ventilated with these settings.
How did they decide which PEEP should be matched with which FiO2? Well, it wasn't exactly scientific. According to the authors, their PEEP/FiO2 protocol represented "a consensus of how the investigators and clinical colleagues balanced beneficial and adverse effects of PEEP in 1995". So, it was based on the current practice of the late 1990s, and its foundations were in the various opinions and beliefs of the ARMA authors. The specific "consensus" they speak of was probably something like the 1996 survey by Carmichae et al. Survey responders reported that "modest levels of positive end-expiratory pressure (PEEP) were used in incremental fashion as FiO2 requirements increased".
Predictably, this approach cannot guarantee success. For instance, Grasso et al (2007) found that this method results in a lot of lung stress and alveolar hyperinflation. The main problem is probably the tendency of the protocol to recommend PEEP which is much lower than one might intuitively use. For example, if your patient only requires 30% FiO2, the protocol recommends a PEEP of 5, which would probably be a sub-optimal level of PEEP for most ARDS patients. The low PEEPs are contrary to the open-lung ethos. Moreover, the higher PEEP recommendations in the high-FiO2 range also fail to take into acount the potentially disastrous shunt effects of such a high PEEP in patients with diminished right ventricular function and pulmonary hypertension (which are not exactly uncommon in the ARDS group).
The PEEP can be adjusted up or down to determine the maximal compliance. The "optimal" PEEP is therefore one at which compliance will be maximal - i.e. increasing the PEEP beyond this magical point will do little to improve the compliance, and decreasing the PEEP below this value will make the compliance worse.
Even though a lot has been published on this topic, a good representative study needs to be selected for the purposes of short pre-exam discussion, and Pintado et al (2013) is as good as any. The authors used an ancient method of determining optimal PEEP from static compliance, based on a 1978 paper by Suter et al.
Pintado et al found this worked for them, in their small RCT of 70 patients. The study lacked power to detect this mortality difference, and so the trend towards improved survival they observed was not statistically significant, even though the effect size was substantial (a 28-day mortality difference of 20.6% vs. 38.9%, which would probably be interesting if it were ever repeated in a large RCT).
Unfortunately, it may all be bullshit. Their comparison group was treated with the ARDSNet protocol, which in most cases resulted in essentially the same mean PEEP settings as the static compliance method. In short, the groups were similar to a frightening degree, which does invalidate the mortality findings to some extent. Overall, 80% of the compliance-method patients received "different PEEP", but it was different by an insignificant value (usually, 1cm H2O higher). The data was interesting enough to further stimulate the ART trial, but not interesting enough to change practice.
One way to find the lower inflection point of the pressure volume curve. The art and science of how to figure out where that inflection point is has been discussed elsewhere.
In order to work properly, this method requires a completely paralysed patient. One may use the lower inflection point (which represents alveolar recruitment), or - perhaps more accurately- one may use the upper inflection point, were derecruitment occurs. In either case, nobody really knows where these points actually are when presented with a real pressure-volume loop. When faced with the challenge of determining an optimal PEEP a selection of clinicans tended to disagree by as much as 11 cmH2O.
Another important limitation of this technique (if being unable to find the inflection point was not enough) is the fact that recruitment of lung units does not uniformly occur at the lower inflection point, but rather all over the presure-volume curve. Some lung units (up to 30% of them in ARDS) may remain closed even at 30cm H2O. Pelosi et al (2001) found that pressures up to 45-60 cmH2O were required to recruit those units.
Another method is to open up all the alveoli with a recruitment manoeuvre, and then to “work backwards” from a high pressure, watching the SpO2 each time you make an incremental downward change in PEEP. Once the sats drop by more than 2%, one can conclude that some sort of derecruitment has taken place, and put the PEEP back up by 2cmH2O.
One may recruit with stepwise increases of PEEP, or one may de-recruit with stepwise decreases. Both methods seem equally effective (or equally ineffective) in determining the optimal PEEP. The major advantage of this method is its simplicity (the only resource you need is time, as the patient is already having their saturation monitored). The pragmatic nature of the endpoint is also a bonus: the oxygen saturation is an important parameter to aim or when ventilating an ARDS patient.
Unfortunately, no "standard" staircase protocol is available. People cannot agree on the ideal recruitment manoeuvre, the ideal starting PEEP, magnitude of the decrement, or whther to measure arterial oxygen tension, saturation, shunt, or whatever. But we need something, some sort of protocol to cling to in this sea of uncertainty. A reasonable practical guide is offered in a good paper by Suh et al (2003). The authors used the following strategy:
This is probably no better or worse than any other such strategy. Another variant is promoted by Tuxen and Hodgson, from the the Alfred. They recruited up as well as down, and used SaO2 instead of PaO2. The details are described in the protocol of the PHARLAP trial (2010) as well as this 2011 paper. Twenty ARDS patients seemed to benefit, but the outcome measures were far from patient-centred ( plasma IL-8 and TNF-α levels).
Recall the shunt equation, discussed in greater detail elsewhere:
By this calculation, one is able to calculate the volume of blood which passes through the derecruited pus-filled lung, bypassing gas exchange.
Theoretically, as one recruits more lung, the shunt fraction should improve- up to a point- and then deteriorate, as excessive intrathoracic pressure increases shunting of blood out of ventilated areas.
A good example of this being put to use is in a 2011 article by Mahmoud et al. The authors were using the (Qs/Qt) to compare between two different recruitment manoeuvre strategies. The technique is thoroughly old-school, described since the 1970s (eg. Suter et al, 1975)
The ARDS patient will have a widended gradient between their arterial and their end-tidal CO2, because of the "wasted" dead space ventilation. So little lung tissue is actually exchanging gas that even CO2 cannot escape. It is therefore posited that the ideal PEEP is one at which the dead space fraction (VD/VT) is minimal. This is probably also the PEEP at which compliance is maximal. Murray et al (1984) tried this in a canine ARDS model (adult mongrel dogs injected with oleic acid) and found that this variable is even more sensitive than the calculation of intrapulmonary shunt. The obvious advantage of this technique is that it does not require a PA catheter.
Practically speaking, this technique involves taking frequent arterial blood gases (which you are probably already doing). Each gas is then compared to the EtCO2 measurement at the instant of sample collection. The difference between PaCO2 and EtCO2 is then plotted. The relationship of PEEP and PaCO2-EtCO2 will end up looking something like this graph from the Murray paper:
As one can see from this grainy scan, at a certain PEEP the gradient is minimal, and then it climbs again (presumably because the shunt worsens with increasing intrathoracic pressure).
It is possible to determine the ideal PEEP by looking at repeated CT scans of the patients at different pressures. The exposure to radiation alone is enough to make this an undesirable option, let alone the logistics.But put aside the need to regularly have pleasant jaunts to and from the CT scanner with your unstable ARDS patient; does this techique actually work? Turns out, no.
Cressoni et al (2014) subjected 51 patients to this sort of graded-pressure radiation torture, and found that lung recruitability and CT-derived PEEP values were completely unrelated.
The assumption was that PEEP is the pressure required mainly to lift up the lung and anterior chest wall which is compressing other lung, leading to dependent lung collapse. The CT-guided approach rests of the hypothesis that the whole point of PEEP is to overcome this gravity-dependent collapse. Cressoni and coleagues used some basic maths to calculate the amount of PEEP required to lift the chest wall and overcome the compressing hydrostatic pressure in the lung bases, on the basis of two CT scans per patient (one at 5 and one at 45 cmH2O).
The CT images above were not from the Cressoni paper; they are reproduced with no permission whatsoever from Chiumello et al (2016). They demonstrate the difference between recruiters and non-recruiters. Some of the patients responded poorly to a 40 cmH2O pressure challenge, whereas others enjoyed a substantial improvement in their lung aeration.
From the difference in CT scans, the investigators were able to estimate the "optimal" PEEP which should be required to overcome the lung and chest wall weight. The outcome was surprising. They found that CT-derived PEEP values were completely unrelated to the amount of recruitment from a 40cm change in pressure.
"The average PEEP needed in patient with ARDS to keep their lung open is approximately the same (approximately 16 cm H2O) independently on the fact that the amount of tissue to be kept open is as low as 3% or as high 50% of the total lung weight".
That is to say, patients who had virtually no recruitment at 45 cmH2O and patients who recruited very well both required about the same level of end-expiratory pressure to keep their lungs open - around 16 cmH2O. That is probably the main take-home message of the CT-guided approach. The optimal PEEP setting is unrelated to lung recruitability.
This had become a hot topic in 2016, to such an extent that it had merited an entire 10-mark SAQ (Question 25 from the first paper of 2017). This level of attention from the college justified the use of a whole chapter dedicated to the topic, and the key concepts from it will not be overelaborated here. At the most basic level, transpulmonary pressure is the pressure which distends the lungs. It is the difference between the pressure inside the alveoli (conventiently available as Pplat) and pressure inside the pleural cavity (which cannot be conventiently measured, but for which the oesophageal pressure is a reasonable surrogate).
Though this sounds like an excellent idea, it has many limitations. The explanation in the LITFL CCC entry for this topic draws upon an excellent article by Sarge et al (2009). Another recent article (Sahetya et al, 2016) also brought up a nice list of advantages and disadvantages, which sounds like something the college might ask about at some point in the future. The summary below was cobbled together from these sources.
What is the meaning of this variable? How do you use it? Well. In brief, one needs to regularly perform expiratory and inspiratory hold manoeuvres to use the TPP. A low (or even negative) expiratory TPP will lead to derecruitment and atelectasis, whereas a high end-inspiratory TPP will lead to VILI.
TPP has the advantage of separating chest wall compliance from lung compliance. One should theoretically be better able to ventilate the morbidly obese and patients with high abdominal compartment pressure. For instance, in a patient with a massively obese chest wall the pleural pressure may be highly positive. Let's say it is 15 cmH2O. At a Pplat of 30 cmH2O, the TPP is still only 15. With an expiratory hold at a PEEP of 10, the Pplat ends up being 12 cmH2O, giving a TPP of -3 cmH2O. In this scenario, the fat patient develops atelectasis - clearly more PEEP is required.
Variables affecting Pes
Situations in which the Pes does not correlate with pleural pressure
Advantages of TPP-guided PEEP selection
Disadvantages and limitations
The electrical resistance of the thorax should plateau at the point where all the recruitable alveoli are recruited (Long et al, 2015); one may even be able to assess the degree of ventilatory inhomogeneity (and how this changes with recruitment manoeuvres).
The advatgaes of this technique are its non-invasive nature, its freedom from radiation exposure and the logistic convenience (i..e not having to go to and from a CT scanner). Though CT of the lung is said to be the "gold standard" in terms of spatial resolution (i.e. you see all the lung units), EIT is much better at temporal resolution, i.e. the ability to plot lung recruitment vs. time. It offers the opportunity for live dynamic lung imaging in the course of mechanical ventilation.
The most detailed overview of this is available from Kobylianskii et al (2016). They found 67 studies worth reading, and concluded that EIT has been " demonstrated to be useful". However, to date there is no outcomes evidence. The extenral validity of human studies has so far been limited by small sample size.