[See the previous Gotchas here: peri-intubation arrest and pulmonary hypertension]
The patient is a 63-year-old female admitted with severe multi-focal pneumonia. She is intubated for hypoxemia shortly after arrival and placed on typical lung-protective settings.
- Mode: Assist control with a volume target (i.e. volume control)
- Rate: 12
- Tidal volume: 450 ml
- PEEP: 5 cmH2O
- FiO2: 40%
Hypoxemia persists after intubation, and you rapidly titrate up her FiO2 to 100%. Using a PEEP-to-FiO2 scale similar to the ARDSnet protocol, you simultaneously increase her PEEP. However, oxygenation continues to be poor, and before long you have a saturation of 70% and PaO2 of 40, despite a PEEP of 25. You are contemplating proning and ECMO when an experienced respiratory therapist suggests reducing the PEEP.
Warily, you drop the PEEP to 20, then 15. With each decrease, you are puzzled to see your saturation noticeably and rapidly improve. Soon, you’re at a PEEP of 8, with a saturation of 100%, and are actually able to reduce your FiO2 to 60%.
What happened?
Why do we PEEP?
Although knee-jerk responses like “increased hypoxia = increased PEEP” have their role, it’s always a good idea to remember the underlying physiology while you’re twiddling the vent knobs.
The fundamental reason for PEEP is to maintain alveolar recruitment. As positive pressure flows into the lung, the alveoli expand; as pressure flows out, they contract, and due to their native elastance, will tend to collapse completely at end expiration when the pressure drops to zero. This is prevented in normal lungs by surfactant (which defrays surface tension and hence reduces the collapsing pressure) and to a degree by autoPEEP. In diseased, intubated lungs, however, these processes are impaired. Combine that with localized airspace processes like pneumonia or pulmonary edema, and you end up with lung units that are atelectatic (collapsed) or filled with something other than air (consolidated).
With the positive pressure that occurs during inspiration, you may be able to recruit some of those deaerated alveoli. However, as the pressure abates during expiration, they will immediately derecruit. This series of inflate-collapse cycles is harmful, since each instance of “popping” open and slamming shut causes microfractures in the alveoli (atelectrauma) and consequent lung injury.
So, we add PEEP, maintaining a positive airway pressure and volume at all phases of the respiratory cycle. This attenuates lung injury, but also facilitates recruitment, a product of not only pressure but time. Each breath recruits some alveoli, which are held open by PEEP, allowing the next breath to recruit more and more, producing gradual stepwise progress toward fully open lungs, instead of a “reset” at the end of each breath.
As a rule, the pressure required to recruit is always higher than the pressure needed to maintain that recruitment: you might need 30 cmH2O to pop an alveolus open, but only 6 cmH2O to keep it open. Hence, normal ventilation (or a brief recruitment maneuver at supra-normal pressures) establishes recruitment, while PEEP maintains it.
The trouble with shunt
The real challenge is that sick lungs are not made up of a single alveoli, or even a collection of identical alveoli. They are heterogeneous.
Imagine a two-unit lung with one open, recruited alveolus, and one shunted, collapsed alveolus.
Notice that here, all of the delivered pressure and volume are going preferentially to the open left alveolus. Not only is this pointless, it could result in lung injury, particularly if you calculated your ventilator settings with the expectation of a whole lung to accept each breath. This is the real reason that ARDS patients need smaller-than-normal breaths: the volume of functional lung is simply shrunken, so patients need to be ventilated like babies (“bambino lungs” in the words of Gattinoni).
Now, your instinct will be to try and pop open that alveolus on the right, and perhaps you can. If the cause of shunt is simply a heavy chest wall or a thin, transudative pulmonary edema, then increasing the delivered pressure and volume may successfully recruit the collapsed lung and keep it open.
But suppose the shunt is caused by a thick, purulent, proteinaceous exudate from pneumonia or ARDS. Or clotted blood from massive hemoptysis. Or even mechanical obstruction from a malignant mass. That portion of lung may be unrecruitable, at least by any reasonable, physiologic amount of pressure. You’re just not getting any air in there, period.
If you try in vain to recruit those densely consolidated, unrecruitable lung units, what will happen? Well, you’ll continue to increase the pressure and volume delivered to the lungs. For most clinicians, this means increasing the PEEP, perhaps with the addition of recruitment maneuvers (although these have somewhat dwindled in popularity). However, none of this added PEEP will reach the bad lung. Instead, it will preferentially enter the “good” alveoli, and increasingly overdistend it.
This will result in volutrauma of the healthy lung, perhaps leading to further lung injury in the middle-to-long term. But in the short term, could it actually cause hypoxia?
Worsening V/Q matching caused by overdistention
Turn your attention now to the pulmonary capillaries.
These poor fellows were innocent bystanders until now, doing their job as expected. In the shunted alveolus, they were vasoconstricting to redirect blood toward the good side, limited the blood that passes uselessly through the shunted lung and maximizing the perfusion to the ventilated region.
However, like the rest of the pulmonary circulation, these are low-pressure vessels. It doesn’t take much to compress them.
You can probably see where this is going.
As you increase PEEP fruitlessly, and increasingly distend your working, ventilated alveoli, that alveolar overpressure will begin to collapse the pulmonary capillaries within the recruited lung. As they’re squeezed shut, perfusion to ventilated lung is decreased, and perfusion to the shunted lung is consequently increased. Voila: you’ve worsened V/Q matching and hence exacerbated hypoxemia.
How can you avoid it?
It’s easier than you might think to get caught in this trap.
Increasing PEEP alongside FiO2 is a reasonable starting point. But it’s always important to bear in mind that some lung is not recruitable using any amount of PEEP. Unlike our two-unit model, of course, real lungs exist on a spectrum, with some freely ventilated alveoli, some that are densely consolidated, and some in the middle that may be recruitable. You can’t always quantify this beforehand. But you can appreciate the end result, which is that some patients will respond well to increased PEEP, while others may not.
So turn up your PEEP—and perform a recruitment maneuver if that’s your thing—then allow some time for recruitment to occur. If, however, there is no improvement in the oxygen saturation after a reasonable period (perhaps 15 minutes), then you have probably failed to recruit the problem lung units. You can either try more PEEP, give it more time… or call it quits, and stop trying to open those alveoli until the underlying disease has begun to improve.
(Measuring driving pressures before and after PEEP changes may be a useful way to quantify this process, since the driving pressure should decrease after an effective PEEP increase and increase after a fruitless one.)
At the other end, if you have a sick, hypoxic patient on very high levels of PEEP (or perhaps APRV), always consider whether you may have gotten ahead of yourself on the PEEP rollercoaster, and consider a trial of reducing it. This should be done cautiously, since lung recruitment is a precious commodity in badly shunted patients that—once lost—may take hours to rebuild. But if you’re scratching your head over a patient with relatively unimpressive markers of disease (such as chest imaging), yet severe hypoxia and high PEEP, it’s a move to try.
Like most things in critical care, there’s always a right amount of PEEP—you just may not know what it is. But if you remember that the dial turns in both directions, you’ll have a better chance of finding the sweet spot.
[See the next Gotcha here: Peak pressure limits]