Show Summary Details
Page of

Ventilator trauma in the critically ill 

Ventilator trauma in the critically ill
Ventilator trauma in the critically ill

Marcelo Amato

and Andreas Wolfgang Reske

Page of

PRINTED FROM OXFORD MEDICINE ONLINE ( © Oxford University Press, 2021. All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in Oxford Medicine Online for personal use (for details see Privacy Policy and Legal Notice).

date: 26 January 2021

Key points

  • Ventilator trauma refers to complications of mechanical ventilation that have an impact on morbidity and mortality.

  • Two major forms of ventilator trauma may be distinguished, an acute form related to rupture of airspaces causing air leak syndrome and a subacute form causing protracted inflammatory responses.

  • The most relevant consequence of airspace rupture is tension pneumothorax, which requires emergency intervention to prevent respiratory or haemodynamic collapse.

  • Subacute ventilator trauma results from cyclic sublethal cell injury during mechanical ventilation and triggers intense, potentially self-perpetuating inflammatory responses.

  • A marker-variable to monitor the risk of subacute ventilator trauma is driving-pressure. Because it is closely related to lung compliance and thus to the size of the lung compartment that remains functional, driving-pressure can aid in identifying disproportionate combinations of tidal volumes and lung compliance.


Two major forms of ventilator trauma, occurring as complications of mechanical ventilation, can be conceptualized. The first form is the occurrence of extra-alveolar gas, also called gross barotrauma or air leak syndromes. Most air leak syndromes, such as pneumothorax or subcutaneous emphysema, cause acute problems. In contrast, the second form of ventilator trauma represents a subacute problem that can be easily obscured by complex critical illness. Except for extreme conditions found in experimental models, subacute ventilator trauma often causes protracted lung inflammation, fibrosis, and deterioration of lung function, contributing to systemic inflammation and organ failure. Clinical distinction of the two forms of ventilator trauma is not always straightforward. For example, gross barotrauma may cause venous gas embolism, leading to pulmonary hypertension, inflammation, and deterioration of gas exchange, ultimately mimicking the clinical picture of the subacute form. The recognition that subacute ventilator trauma may impair prognosis and increase mortality in mechanically-ventilated patients was one of most important discoveries of the last decades, and directly affected our daily practice of mechanical ventilation.

Gross barotrauma

Extra-alveolar gas usually escapes through ruptured alveoli or terminal airways at the junction with the vascular sheaths, due to large pressure gradients between intra-alveolar and interstitial spaces. Large pressure-gradients between these two regions are created not only during positive-pressure ventilation, but also during strong spontaneous efforts when very negative interstitial pressures are created. Gas accumulation in the pulmonary interstitium is called pulmonary interstitial emphysema. Gas may migrate through the interstitium, either peripherally, forming subpleural cysts, or centrally reaching the hilum and mediastinum. Pressurized mediastinal or subpleural gas collections can rupture and cause pneumothorax. Many pneumothoraces occurring during positive-pressure ventilation are under tension, necessitating prompt intervention to prevent respiratory or haemodynamic collapse. Gas can also spread along fascial planes to the neck, thorax, or abdomen, and cause subcutaneous emphysema or pneumoperitoneum.

The current incidence of gross barotrauma during a prolonged course of mechanical ventilation is below 10%, much lower than the figures reported before the advent of lung-protective mechanical ventilation. Although gross barotrauma is rarely the cause of death, its occurrence indicates severe underlying lung pathology, and is statistically correlated with increased morbidity and longer intensive care unit (ICU) stay. Common risk conditions for gross barotrauma are acute respiratory distress syndrome (ARDS), pneumonia, aspiration, and air trapping due to obstructive airway diseases. Such conditions are commonly associated with global airspace overstretching, as in asthma, or with localized overstretching, as in ARDS, where few lung regions remaining aerated (the ‘baby lung’) must accommodate unevenly distributed tidal volumes [1]‌.

Interestingly, gross barotrauma has not been necessarily associated with high positive end-expiratory pressure (PEEP), except for old studies using large tidal-volumes. In recent clinical studies limiting airway plateau-pressures below 35 cmH2O, no clear association between high airway pressures and barotrauma could be detected.

Ventilator-induced lung injury

The subacute form of ventilator trauma, named ‘ventilator-induced lung injury’ (VILI) in experimental studies or, more conservatively (when the relative contributions of mechanical ventilation and underlying disease are less clear), ‘ventilator-associated lung injury’ is histologically indistinguishable from the diffuse alveolar damage associated with ARDS. In lung samples from experimental animals and patients, light microscopy revealed alveolar haemorrhage, neutrophil infiltration, hyaline membranes, intra-alveolar and interstitial oedema, increased septum thickness, airspace collapse, and proliferation of macrophages and type II pneumocytes. Within terminal airways, alterations such as intraluminal inflammation, epithelial denudation, and necrosis have been described. Later, in a more chronic phase, alterations include bronchopulmonary dysplasia, cysts (dilated airways), airway remodelling, and fibroproliferation. Electron microscopy further reveals endothelial/epithelial abnormalities and tears, disruptions of the basement membrane, formation of intracapillary blebs, and an abnormal surfactant layer.

A key feature of mechanically-ventilated lungs, whether normal (during anaesthesia ventilation) or diseased, is the presence of non-aerated and unstable regions due to atelectasis, oedema or consolidation. Besides representing the disease itself, loss of aeration has been proposed as a central biophysical promoter of VILI. It was shown that, in the same lung, unstable lung regions were injured by VILI, whereas stable regions were not. The classical concept of mechanical interdependence suggests that pressures acting locally in non-uniformly expanded lungs at the boundaries between the atelectasis and aerated lung may be a multiple of the apparent transpulmonary pressure [2]‌.

Synergistically with but also independent from non-uniform lung aeration, cyclic airspace overstretching, alveolar instability (due to surfactant dysfunction or negative transpulmonary pressures, for instance), and lung inflammation, have all been reported to precipitate or contribute to VILI. The sequelae of these mechanisms and the resulting intensity of VILI seem to depend on the intensity and duration of ventilation and are not restricted to mandatory ventilation. Among the physical mechanisms of VILI is the detachment of epithelial and endothelial cells from basement membranes and rupture of cellular and/or cytoskeletal structures. In addition to physical injury, a generalized pulmonary inflammatory response can cause and maintain VILI. Pulmonary inflammation involves neutrophil accumulation and activation in the pulmonary capillaries, the interstitial, and the alveolar spaces, and potentiates the destabilization of the alveolar–capillary barrier. The consequences are the development and dispersion of oedema, loss and inactivation of surfactant, loss of integrity of the alveolar compartment, and the transfer of bacteria and inflammatory mediators into the systemic circulation [3]‌.

Accordingly, inflammatory cytokine concentrations (e.g. TNF-α‎, interleukin-(IL)-1β‎, IL-6, or IL-8) in broncho-alveolar lavage fluid or serum are associated with more injurious ventilation and with higher morbidity or mortality. Among the mechanisms by which cellular and subcellular pulmonary structures can sense injurious mechanical stimuli, and trigger and maintain local and systemic inflammatory or reparative responses are transformation of strain into (bio)chemical signals (mechanotransduction), stress failure of cytoskeletal structures (necrosis), and impairment of the alveolar barrier (decompartimentalization). These mechanisms can be activated also by unilateral injury or even without ultrastructural damage [3]‌.

Several of the afore-mentioned pathomechanisms have the potential to become self-perpetuating. Proteinaceous oedema, for example, increases the lung weight and inactivates surfactant, which aggravates collapse of the dependent lung regions, worsens the non-uniformity of lung aeration and amplifies the negative effects of mechanical interdependence. The latter in turn alters vascular transmural pressures, promotes the development and distribution of lung oedema, and impairs ventilation-to-perfusion matching. Surfactant depletion and dysfunction, as another example, creates negative perivascular pressures, which together with altered alveolar–capillary barrier permeability, promote the development of oedema.

Seminal studies observed that adding PEEP to injurious mechanical ventilation attenuates VILI and may interrupt the viscious cycle described in the previous paragraph. The generalized readout of such experiments is that PEEP attenuates VILI when the end-inspiratory volume or plateau pressures are maintained unchanged [4,5]. To explain such beneficial effects, many facets of the application of PEEP need to be considered. First, adding PEEP to mechanical ventilation with unchanged plateau pressure reduces the amplitude of both pressure and cyclic (over)stretching. Applying PEEP to a non-uniformly expanded lung reduces the amount of non-aerated lung units, stabilizes unstable lung units, and reduces both non-uniformity and oedema.

Effects on remote organs

The clinical consequences of VILI are not limited to the lung. Injurious ventilation can have deleterious systemic effects. Accordingly, the most common cause of death in ARDS patients is not intractable hypoxaemia, but a systemic inflammatory response syndrome and multiple organ dysfunction. Interactions between local pulmonary, systemic and distal organ inflammation may help explaining the decreases in serum levels of inflammatory cytokines, associated with decreased morbidity and mortality observed in recent trials of lung-protective mechanical ventilation [6]‌.

Barotrauma versus volutrauma

After the recognition of subacute ventilator trauma unrelated to air leaks, it was proposed to replace the old term barotrauma (i.e. pressure-induced trauma) by volutrauma (volume-induced trauma). The rationale for this new terminology came from two remarkable observations:

  • Chest strapping, by significantly decreasing chest-wall compliance, minimized VILI despite the application of very high positive airway pressures.

  • Negative pressure ventilation, by cyclic application of vacuum around the chest wall, could produce severe VILI—despite atmospheric airway pressure.

Tidal volumes generated in the first experiment were small, but very high in the second experiment. Thus, injury was not solely related to high airway pressures—it only occurred when high tidal volumes were concomitantly generated [4]‌.

Although representing an improvement, the term volutrauma was later shown to be also imprecise. It became evident that high tidal volumes are not always injurious. Conversely, in lungs with gross collapse and low tidal-compliance, VILI can occur despite ‘protective’ tidal volumes (as low as 4 mL/kg). To better estimate injury risks, it was later proposed that tidal volumes should be individualized according to both the potential size of the whole lung (normalized by height or ideal body weight), but also according to the size of the functional lung compartment (spared from disease) that accommodates the tidal volume. This reduced functional lung volume in ARDS patients has been termed the baby lung [1]‌. The lung areas within the baby lung that are spared from collapse, are particularly exposed to VILI because they are exposed to significantly increased mechanical work. Since shunt makes gas exchange less efficient, the spared zones have to achieve the whole ventilation task. Studies using positron emission tomography have shown that the most inflamed areas in injured, mechanically-ventilated lungs are the ones receiving the highest ventilation, and not the silent, non-aerated regions.

Stress, strain and cyclic lung stress

Trying to circumvent the main limitations of the terms barotrauma and volutrauma, the engineering terms stress and strain were recently proposed for defining risk-constellations for VILI.

Stress represents an evolution of barotrauma. Risk is now defined in proportion to transpulmonary pressures (airway minus pleural pressure) and it becomes possible to explain the attenuation of VILI by chest strapping. It is also possible to explain the pronounced VILI associated with negative pressure or vigorous spontaneous ventilation—strongly negative pleural pressures may generate high transpulmonary pressures even against atmospheric airway pressure.

Conversely, strain represents an evolution of volutrauma and risk is now defined in proportion to resting lung volume. Strain equals the ratio between end-inspiratory lung volume and functional residual capacity (when central airways are submitted to atmospheric pressure). Thus, it can be explained why even small tidal volumes can be injurious in patients with gross lung collapse—the resting lung volume (the denominator) decreases and strain becomes high.

One main limitation of the stress/strain approach is its physical reductionism: it assumes that stress on the structure can be estimated as a quasi-linear function of strain. In fact, a complex structure like the lung, composed of different cell types attached to a network of extracellular matrix, elastic and collagen fibres, reacts to physical deformation in a markedly non-linear manner. For instance, the response of pneumocytes to surface deformations depends on their repair capabilities. There is a clear threshold, beyond which further strain causes cell wounds and uncontrolled transmembrane ion traffic. The lung has a very large reserve of alveolar surface, which is folded during resting conditions. Therefore, the delicate lung surface will be stressed only when unfolding is no longer possible. Below this threshold, prolonged large variations in surface area (i.e. during a marathon race) are tolerated without significant injury. Assuming that this non-linearity is constitutive for lung function, the interchangeable use of strain/stress becomes troublesome. The assessment of stress requires oesophageal pressure measurement as a surrogate of pleural pressure, which is rarely reliable in supine patients with massive atelectasis. Proper strain calculation requires whole-lung computed tomography and many approximations for the calculus of recruitment and true ‘surface deformation’ (as opposed to surface unfolding). In conclusion, the actual stress will hardly be represented by in vivo strain measurements, especially when considering measurement errors at the bedside.

Another limitation of the stress/strain engineering perspective is the omission of consideration of the ‘pace of deformation’—vital cells easily adapt to sustained deformations by the neoformation of cell membranes, but they rarely survive rapid, large changes of alveolar surface area. It has been shown that strain preconditioning helps cells to stand a subsequent large deformation [3]‌. Translated to a larger scale, increasing end-expiratory volumes by application of PEEP may have preconditioning effects protecting against injury by high inspiratory lung volumes (strain). This would be another explanation for protective effects of PEEP.

Bedside estimate of cyclic lung stress

Based on the principles exposed previously, absolute end-inspiratory thresholds should be de-emphasized for assessing the risk of VILI. Cyclic deformation-thresholds should be proposed instead. Variables indicating cyclic stress variations (pressure swings) applied to the lung tissue seem preferable to markers of area/volume deformation, since the latter are difficult to assess at the bedside. Transpulmonary pressures should be included in the consideration of ventilator settings: increased pleural pressures might counterbalance stress, while negative pleural pressures may amplify stress.

A marker-variable fulfilling most of the aforementioned characteristics is driving-pressure [7]‌. During controlled mechanical ventilation, driving-pressure is the difference between end-inspiratory and end-expiratory pressures, both measured during the brief pauses while flow is zero. It represents the total cyclic pressure gradient applied across the respiratory system for elastic expansion of lung plus chest wall. Thus, part of the total driving-pressure applied to the respiratory system is counteracted by chest wall elastance, which means that cyclic lung stress would be better represented by isolated measurements of transpulmonary pressure. Nevertheless, the simplification of using driving-pressure as a marker-variable is convenient because it renders estimation of pleural pressure unnecessary and it is also reasonable because lung elastance during acute diseases is typically 5–10 times higher than chest wall elastance [7]. Therefore, the majority (80–90%) of driving-pressure is indeed applied to the lung. Moreover, obesity and intra-abdominal hypertension usually cause positive offsets in pleural pressures, but rarely increase chest-wall elastance, which is relatively constant among patients.

Assuming that driving-pressure mirrors cyclic changes in transpulmonary pressure shows why this variable has the best predictive power for VILI in experimental and human studies, and outperforms simple measurements of plateau pressure or tidal volume [7]‌. Moreover, driving-pressure can be calculated as the ratio between tidal volume and compliance. Because the size of the baby lung is positively correlated to compliance, driving-pressure will indirectly account for the size of the baby lung—for a given tidal volume [1].

For application of driving-pressure as a surrogate of cyclic stress, a driving-pressure threshold would aid in limiting cyclic deformation and VILI. Obviously, no well-established threshold can yet be proposed. However, in all recent trials testing lung-protective ventilation strategies, reduction of driving-pressure (usually to below 15 cmH2O) was associated with reduced morbidity and mortality [7]‌. This driving-pressure threshold is also supported by millions of patients who undergo uncomplicated anaesthesia ventilation, where driving-pressure is commonly below 15 cmH2O.

The driving-pressure concept can also be applied in the presence of spontaneous breathing efforts when some reasonable assumptions are made. If the respiratory system compliance can be measured during a preceding period of controlled ventilation, the tidal volume during subsequent assisted spontaneous efforts allows estimation of the total driving-pressure applied to the system by ventilator and inspiratory muscles. Using this approach, potentially injurious combinations of compliance and tidal volume can be identified. For instance, a tidal volume of 500 mL measured during low levels (e.g. 5 cmH2O) of pressure support, together with a recent compliance measurement of 20 mL/cmH2O, indicates that a very high total driving-pressure of 25 cmH2O is applied, deserving the same concern as if it was observed during controlled ventilation.


Still today, acute and subacute ventilator trauma cause significant extra-morbidity and -mortality in mechanically-ventilated patients. The traditional concepts of barotrauma and volutrauma evolved to more sophisticated considerations about physical stress and strain applied to the delicate lung tissue. By representing integrated information about cyclic lung stress and specific responses of the lung tissue, driving-pressure can work as a marker-variable to predict the risks of VILI at the bedside.


1. Gattinoni L and Pesenti A. (2005). The concept of ‘baby lung’. Intensive Care Medicine, 31, 776–84.Find this resource:

2. Mead J, Takishima T, and Leith D. (1970). Stress distribution in lungs: a model of pulmonary elasticity. Journal of Applied Physiology, 28, 596–608.Find this resource:

3. Plataki M and Hubmayr RD. (2010). The physical basis of ventilator-induced lung injury. Expert Reviews in Respiratory Medicine, 4, 373–85.Find this resource:

4. Dreyfuss D, Soler P, Basset G, and Saumon G. (1988). High inflation pressure pulmonary edema. Respective effects of high airway pressure, high tidal volume, and positive end-expiratory pressure. American Reviews of Respiratory Diseases, 137, 1159–64.Find this resource:

5. Webb HH and Tierney DF. (1974). Experimental pulmonary edema due to intermittent positive pressure ventilation with high inflation pressures. Protection by positive end-expiratory pressure. American Reviews of Respiratory Diseases, 110, 556–65.Find this resource:

6. Slutsky AS and Imai Y. (2003). Ventilator-induced lung injury, cytokines, PEEP, and mortality: implications for practice and for clinical trials. Intensive Care Medicine, 29, 1218–21.Find this resource:

7. Amato MBP, Meade MO, Slutsky AS, et al. (2015). Driving pressure and survival in the acute respiratory distress syndrome. New England Journal of Medicine, 372, 747–55.Find this resource: