Patient self-inflicted lung injury (P-SILI)

Patient self-inflicted lung injury is an interesting variation of victim-shaming, which accuses the patient of accelerating their own ventilator-associated lung trauma, while also blaming the intensivist for letting them get away with it. The concept, though ancient, has reappeared in recent years with the appearance of COVID19, which produced a phenotypically abnormal ARDS featuring good compliance in spite of terrible gas exchange. These patients would often be seen to take massive tidal volumes, driven by their air hunger and pulmonary irritation. And the intensivists looked upon this, and they said, lo; it profanes God's name to breathe in this unnatural manner.

In summary, if anyone were ever asked to discuss this in any sort of detail,

Definition

• P-SILI is the concept that lung injury could potentially be induced or worsened by the patient’s own spontaneous inspiratory effort.

Rationale and physiological explanation

• The stress (pressure) and strain (volume change) generated during ventilation is what does the mechanical damage of VILI
• High strain and stress are thought to be harmful to lung tissue; protective low tidal volume mechanical ventilation is thought to improve mortality from moderate and severe ARDS by reducing the strain and stress on lung tissue.
• There is no difference between the transpulmonary pressure or lung strain produced by pressure originating from the ventilator, as compared to the pressure originating from the patients’ respiratory muscles
• Ergo, increased patient respiratory effort that uses large amounts of force to generate large tidal volumes should be equally injurious.

Preventative strategies to ameliorate P-SILI

• Reduce respiratory effort
  • Reduce O₂ consumption and CO₂ production by treating fever, pain and agitation.
  • HFNP to improve the efficiency of ventilation
  • Reduce the dead space
  • Opioids to reduce the central medullary sensitivity to hypoxia and hypercapnia
  • Avoid permissive hypercapnia
  • Correct metabolic acidosis
  • ECMO or ECCO₂R 
• Distribute the lung strain more homogeneously:
  • Prone ventilation
  • High PEEP
• Decrease the ventilator contribution to P-SILI
  • Manage dyssynchrony
  • Adjust the cycle-off variable and pressure support to keep the tidal volume in the 6-8 ml/kg range
• Decrease diaphragmatic activity
  • Sedation and paralysis, or partial neuromuscular blockade

Advantages of changing management to prevent P-SILI

• ARDS of all causes is associated with significant mortality
• The use of NMJ blocking agents which may be required to protect patients from P-SILI has not been demonstrated to worsen this mortality
• Respiratory disease that generates the sort of lung stress that risks P-SILI will often require invasive mechanical ventilation, and early intubation may have a protective effect

Disadvantages of changing management to prevent P-SILI

• The risk from P-SILI is difficult to estimate, or to separate P-SILI from the VILI that develops in lung disease that is susceptible to P-SILI
• Respiratory effort is difficult to objectively measure
• There is no available ventilator strategy that is known to specifically reduce the risk of P-SILI
• The pursuit of lower respiratory effort may result in some spontaneously breathing patients requiring intubation
• Targeting lower tidal volumes in spontaneously breathing mechanically ventilated patients may require increased sedation, muscle relaxant use, and the use of mandatory modes of ventilation, which tend to increase the morbidity od critical illness and delay extubation

Own practice

• Anything sane, eg. “I will allow patients to breathe spontaneously as soon as able, without using additional sedation to reduce their tidal volumes. P-SILI remains an interesting hypothetical entity but there is insufficient evidence to change my practice”

It was unexpected that CICM could ever introduce this in their papers, as this is not a widely accepted concept, and to ask questions about something that is still in the realm of "animal data and physiological conjecture" would be contrary to the spirit of the exam process. However, those who are reassured by this statement are redirected to the past paper question about the endothelial glycocalyx; and indeed PSILI did appear in Question 8 from the first paper of 2025. The reader is left to make of this what they will, considering that the interest in this novel concept is waning, as it was really generated mainly by the extremely atypical ARDS experience afforded to the critical care community by COVID19. As such, the reader mostly interested in preparing for realistic exam questions is offered the succinct LITFL article on P-SILI and spontaneous ventilation instead.

Definition of P-SILI

Though most intensivists have now largely converted to ventilating patients with nice low lung-protective tidal volumes, the patients themselves have not (clearly they have not kept up with the literature). And they end up in charge of their ventilation earlier and earlier these days, as the belief that spontaneous modes are beneficial in ARDS has spreads across the critical care community. Modes that allow the patient to have governance over their respiratory mechanics (including APRV) are increasingly common. For example Van Haren et al (2019), combing through the LUNG SAFE data, found that 58% of ARDS patients were breathing spontaneously within the first two days of their intubation. This means the patients are, these days, more often then not, free to inflict untold violence on their own lungs, with the intensivist reduced to watching in powerless horror.  

It appears that there is a lack of standards in naming and defining this phenomenon which has left eminent authors free to refer to it as self-inflicted, self-induced, and effort-dependent lung injury, but all descriptions fortunately share some unifying features:

• The patient needs to be breathing spontaneously (in fact mechanical ventilation is not an essential factor, i.e. the spontaneously breathing patient in the ED waiting room could be accused of this)
• The work of breathing needs to be "excessive", or the effort of breathing has to be "intense". The term "work" suggests that this is something we could measure in joules, and therefore could have a threshold value for, but no such value has been broadly agreed upon.

The CICM exam candidate looking for a one-liner to start their "critically evaluate" answer would probably be not too far from the truth if they were to propose something like:

"Patient self-inflicted lung injury (P-SILI) is the exacerbation of lung injury produced by abnormally elevated transpulmonary pressures generated by the spontaneous breathing efforts of dyspnoeic patients."

Let us explore exactly how this is supposed to happen.

Rationale and physiological explanation

At a basic level, the explanation for P-SILI could be oversimplified as follows:

• Physical forces acting on the lung can produce injury by damaging cells and basement membrane structures, leading to a cycle of reduced compliance that leads to increased distending forces and more damage.
• From the perspective of the lung, it does not matter whether the forces are produced by the mechanical ventilator, the patient, or some combination of the two, as long as these produce forces of the same magnitude acting in the same direction. 
• We know that ventilator-induced lung injury is caused by mechanical stress on the lung due to the actions of the ventilator, and from the above, it follows that mechanical stress caused by the patient's own respiratory efforts should produce the same sort of injury.

Is this for real, one might wonder, looking back on decades of standard ICU teaching. Is spontaneous breathing not supposed to be good for you? Well, reader, it is not the intention of this page to entirely discourage the use of spontaneous ventilation, but rather to introduce into the conversation a possibility that that there might be a range of breathing patterns and work/force/power variables where spontaneous breathing efforts might become harmful. The data to support this had originally arrived in the form of a sheep model from Mascheroni et al (1988). The investigators injected repeated doses of 200mg of sodium salicylate directly into the CSF of sheep, quadrupling their minute volume for the duration of about twelve hours. At the end of the experiment the hyperventilating sheep were significantly more hypoxic, with radiological evidence of lung damage and gross pathological changes suggestive of lung injury. A whole host of similar studies had followed, including those performed on animals with varying levels of lung damage, demonstrating that the most severe ARDS seems to be associated with the greatest vulnerability to P-SILI. 

Pathophysiology of P-SILI

To summarise the events and mechanisms, 

• Increased transpulmonary pressure is thought to be a central concept for mechanical lung injury, where:
  
  Transpulmonary pressure =  (alveolar pressure - pleural pressure)
  
  Generally the trasnpulmonary pressure during end-expiration should be under 10, and during inspiration, under 25. 
• The increased transpulmonary pressure increases the tidal volume, which, whenever anyone is talking about P-SILI, seems to be referred to as lung strain, the ratio of the change in lung volume to the resting lung volume during respiration (dV/V₀). This variable should be below 1.5 for safe ventilation (whereas anything greater than 2.5-3.0 seems to be harmful). 
• The reason "strain" instead of "tidal volume" is used is because the distribution of strain in heterogeneous diseased lungs is very inhomogeneous between different lung regions, i.e. the good lung gets all the volume and, therefore, all the strain.

How is any of this unique to spontaneous breathing, one might ask. It all sounds like something you could also do with positive pressure ventilation. Well:

•  In spontaneous breathing, the negative pressure applied to dependent basal lung regions results in pendelluft, where gas from non-dependent apical regions is redistributed to the bases. This produces overdistension of those regions (Cornejo et al, 2022).
•  The addition of pressure support could theoretically exacerbate the increase in transpulmonary pressure (i.e. with the pressure in the pleura so negative, you have now made the alveolar pressure more positive).
• On top of this, the dyspnoeic patient can develop dyssynchrony with the ventilator, and be injured even further by the erratic pressure spikes and volume fluctuations.
• There is an escalating cyclical pattern, where the more the lung is injured, the more the tissue irritation and gas exchange failure activate the respiratory drive, promoting ever increasing injury.

Also, though not really related to lung injury, we need to consider these factors:

• Increased venous return due to increased negative pressure creates conditions of increased capillary hydrostatic pressure, which favours the formation of pulmonary oedema
• Excessive diaphragmatic effort causes the diaphragm to experience "load-induced injury", similar to what happens in skeletal muscle undergoing intense exercise. That the diaphragm is susceptible to this has been demonstrated by studies such as Jiang et al (1998). 

Differences between P-SILI and VILI

Is there any difference in the injuries sustained during spontaneous ventilation, as compared to mandatory mechanical ventilation? Even though the forces acting on the lungs are theoretically the same, one might expect some difference in the way the two conditions manifest, because the forces may be distributed to different structures. Indeed, data suggest that P-SILI is somewhat less injurious than VILI, at least on an animal model level, where histology of excised lungs seems to demonstrate slightly less damage.  Cruces et al (2023) observed less alveolar and airway damage in the P-SILIed rat lungs, but more damage to vascular structures (perivascular oedema and hyperaemia). The authors speculated that this is because spontaneous breathing creates increased pulmonary blood flow due to negative intrathoracic pressures. Their critics pointed out a series of methodological flaws, but the findings remain interesting.

Patients at risk of P-SILI

Following from the observation that increased effort and poor lung compliance create the environment for P-SILI, it stand to reason that patients at lower risk of P-SILI would be those who are unable or unwilling to generate large swings of transpulmonary pressure, whether due to their weakness or due to the fact that their lungs are in unusually good condition. Thus, the patients most at risk of self-inflicted lung injury are those that:

• Are early in their ICU stay (i.e. still have good muscle strength)
• On NIV, which makes controlling their tidal volumes more difficult
• Or breathing unassisted, which makes it easier to miss large tidal volumes
• Have heterogeneous lung disease, which permits pendelluft and regional amplification of transpulmonary pressure

Identification of P-SILI

How can one tell that P-SILI is occurring? There is no biomarker of lung damage in common use, but there are various parameters one may look to:

• Tidal volume: if the patient is generating giant ones, their lung may be getting overdistended. In the FLORALI trial, where HFNP looked so much better than NIV, it was because the patients in the NIV arm who had tidal volumes greater than 9ml/kg all got intubated and had worse outcomes. 
• Transoesophageal manometry: the only convenient way to measure pleural pressure; this may give you an indication that the transpulmonary pressure is increased. Anything more than 25 during inspiration or 10 during expiration is dangerous.
• Calculate the power of ventilation. This can be approximated from the equation, 
  
  Power = respiratory rate  × (VT × (PEEP + ΔPinsp))
  
  Power of normal spontaneous breathing is about 2-3J/min, and power of mechanical ventilation in ARDS is usually 10-15 J/min; where anything in the 12-17 J/min range is thought to be enough to cause VILI or P-SILI.

Sklienka et al (2023) give a whole range of other options for measuring or assessing respiratory effort, ranging from subjective scales to diaphragmatic ultrasound and EMG, and the reader is left to decide whether their practice will be enriched by reading these. 

Strategies to prevent P-SILI

So, let us consider the scenario where the at-risk population has been confidently identified and the concept of P-SILI adopted sufficiently broadly that targeted strategies can be deployed without fear of intercollegiate criticism on the ward round. What might those targeted protective strategies be? Excellent papers by Goligher et al (2020) and Carteaux et al (2021) formed the basis of the following recommendations:

• Reduce respiratory drive:
  • Reduce O₂ consumption and CO₂ production by treating fever, pain and agitation, which should reduce the total ventilatory requirements
  • HFNP: the action of irrigating dead space with fresh gas improves the efficiency of ventilation, which reduces the respiratory effort. This is conceptually plausible but in reality nobody measures the tidal volumes of these patients and so their respiratory mechanics are difficult to estimate or follow chronologically.
  • Reduce the dead space: respiratory effort may be increased because of an increase in West's Zone 1 due to (for example) overenthusiastic diuresis, which might respond to fluids or some kind of pulmonary blood flow reorganisation (eg. inhaled pulmonary vasodilator or prone position)
  • Opioids reduce the central medullary sensitivity to hypoxia and hypercapnia, and moreover they have all kinds of salutary effects on the respiratory drive by managing the anxiety and discomfort of being critically ill. And if you think that seems like cheating, well:
  • Not permitting hypoxia or hypercapnia is contrary to the usual rules of ARDS, but might be applicable here, as those are the dominant drivers of the increased respiratory drive. Along the same lines:
  • Correct metabolic acidosis.
  • ECMO or ECCO₂R  can remove enough CO₂ to permit relatively normal respiratory drive; or at least the contributions of poor gas exchange to the respiratory drive can be completely abolished by extracorporeal gas exchange.
• Distribute the lung strain more homogeneously:
  • Prone ventilation homogenises lung compliance and decreases the asymmetry between the distension of the apices and bases, which allows spontaneous breathing to take place more safely (as the lower lobes are now more compliant and therefore safer to distend).
  • High PEEP: Yoshida et al (2020) determined that a higher PEEP level can ameliorate a lot of the evil effects of P-SILI , mostly by decreasing the excursion of the diaphragm and distributing lung strain more evenly across the tissue.
• Decrease the ventilator contribution to P-SILI
  • Manage dyssynchrony by all the conventional means
  • NAVA could theoretically reduce the added injury from pressure support by down-adjusting this variable whenever the patient decides to make more vigorous efforts
  • Adjust the cycle-off variable to regulate the tidal volume delivered by supported breaths, so that the tidal volumes remain within the 6-8 ml/kg range
• Decrease diaphragmatic activity
  • Sedation and paralysis, if all the other strategies have failed, will ensure a controlled tidal volume and respiratory rate. This needs to be balanced with the harms of being motionless and comatose.
  • Partial neuromuscular blockade is a compromise described by Somhorst et al (2018) in a case report, where the investigators titrated the low-dose rocuronium infusion to target the desired tidal volume, while the patient continued to breathe spontaneously.

Advantages of changing management to prevent P-SILI

To carry on with letting the patient breathe in a destructive way is folie a deux. Consider:

• Many of the strategies we know to be protective are already mainstream  for other reasons (HFNP, high PEEP, prone ventilation, avoidance of dyssynchrony)
• Sedation and opioid analgesia are often routine care in the ICU and may be indicated for other reasons anyway
• The added cost of longer ventilation due to P-SILI offsets the added cost of expensive technologies such as ECMO and ECCO₂R
• Variables that monitor for P-SILI (eg. power of ventilation) are often available on modern ventilators, to say nothing of basic tools like tidal volume and respiratory rate

Disadvantages of changing management to prevent P-SILI

Following from the list of protective strategies above, the patient being maximally protected from P-SILI is prone, paralysed, sedated, and on ECMO. All of those have their cost:

• We neither have a robust definition which we can apply to research this phenomenon, nor agreed-upon diagnostic criteria, nor any randomised controlled trials to demonstrate the effect of the proposed protective strategies
• Respiratory effort is difficult to objectively measure, especially where the patient is breathing unassisted
• There is no available ventilator strategy that is known to specifically reduce the risk of P-SILI, other than what is recommended as standard practice (i.e. low tidal volumes, low driving pressure, high PEEP, etc)
• The pursuit of lower respiratory effort may result in some spontaneously breathing patients requiring intubation, and some patients weaning on spontaneous modes getting re-sedated to manage their tidal volumes
• Even though P-SILI is thought to be highly prevalent, the paralysed arm of the ROSE trial did not have better outcomes, which suggests that even severe ARDS patients do not seem to be bothered overmuch by spontaneous breathing
• No matter the attachment to preventing P-SILI, spontaneous breathing needs to happen at some stage, as the patient eventually needs to get off the ventilator, in which case all ventilator foreplay that delays this transition can be viewed as a waste of time
• Sedation and opioids are independently associated with increased risk of delirium, weakness, longer ventilation and longer ICU stay, i.e. even if sedation-heavy ventilator weaning strategies reduce P-SILI, they still do not actually reduce the burden of intervention on the patient.