Introduction

Esophageal pressure (Pes) measurement is a minimally invasive monitoring method used to assess respiratory mechanics. In acute respiratory failure patients, Pes tracings can be used to individually understand, define and evaluate the pathophysiological mechanisms of respiratory failure, to titrate invasive and non-invasive respiratory and pharmacological support and to follow the patient’s clinical course. Nonetheless, Pes monitoring is still seldom used in the clinical setting; rather, it remains widely seen as a research tool. In an effort to clarify the usefulness of Pes measurement in the clinical setting, we address here the relevant technical, physiological and clinical details of this monitoring method to facilitate a better understanding of the information that can be obtained by bedside Pes measurement. Two recent reviews have dealt with the same topic [1, 2]. In our review we include updated information from the most recent literature and provide clinical threshold values for transpulmonary pressure (PL) as reference to guide ventilation settings. We also include an extensive analysis on both the clinical value of transmural pulmonary vascular pressure measurements and the correlation between Pes and the electrical activity of the diaphragm. A number of case-scenarios in which Pes monitoring could be of help at the bedside are described (Table 1), and a video on the practical use of the technique is provided in the Electronic Supplementary Material (ESM).

Table 1 Case-scenarios in which esophageal and transpulmonary pressure monitoring could be of help at the bedside

Measurement of Pes

The catheter–balloon system: positioning and calibration

The most common technique for measuring Pes employs an esophageal balloon filled with air, connected to a long thin catheter. The first generation of custom-made esophageal balloons was primarily used for research purposes; since then, several types of second-generation balloons have been developed in the last decade which are currently available for bedside clinical use. Esophageal balloon catheters can be connected to dedicated monitoring devices, to accessory pressure ports of mechanical ventilators or to pressure transducers of multi-parametric monitors.

To obtain reliable Pes measurements, the esophageal balloon must be placed in an appropriate position and the balloon inflated with an adequate volume of air. An under-filled balloon does not properly transmit the Pes, whereas an over-filled balloon can over-estimate the pressure [3]. The “optimal” air filling volume depends on the design, dimensions, geometry and material of the esophageal balloon, which in turn affect its mechanical properties. The six most common commercially available esophageal catheters have recently been tested in vitro at an external pressure ranging from 0 to 30 cmH2O. All of the catheters tested were found to accurately measure the surrounding pressure, but the optimal filling volume differed substantially among the catheters [4]. In addition, the minimum volume required for valid measurements was greater than those recommended previously and was dependent on the surrounding pressure [4]. Another study similarly found that the filling volume varied with the tested catheters and that the volume should be increased in the presence of high external pressure [5]. Physicians must account for these technical issues. In clinical practice, a simple method to determine the optimal filling volume is to progressively inflate the balloon, within the catheter-specific adequate range of filling volumes, finally selecting the lowest volume associated with the largest tidal swings (inspiratory to expiratory) of Pes [4, 6].

Method to insert the catheter and validate Pes during mechanical ventilation

A video describing the insertion and interpretation of the Pes is provided in the ESM. After deflating the esophageal balloon and securing it with a three-way stopcock, it is introduced into the esophagus transorally or transnasally. The esophageal catheter is then gently advanced to a depth of about 55 cm to reach the stomach and inflated with the minimum recommended volume [4, 5]. The intragastric position of the catheter is confirmed by a positive pressure deflection during gentle external manual epigastric compression. Subsequently, the catheter is progressively withdrawn into the esophagus. Esophageal positioning is suggested by the appearance of cardiac artifacts on the pressure tracings and by the change from intra-abdominal to intra-thoracic pressure patterns. Placement of the esophageal balloon in the lower two thirds of the intra-thoracic esophagus has been suggested as a way to limit possible pressure artifacts due to non-homogeneous esophageal compression by external structures [7].

When spontaneous inspiratory efforts are present, the conventional test to validate the Pes measurement is to compare the simultaneous negative deflections of airway and esophageal pressures during an end-expiratory occlusion maneuver (the so-called Baydur test) [3, 8]. During the occluded inspiration, the change in airway (∆Paw) and esophageal (∆Pes) pressure should be almost identical because there is no change in lung volume and, thus, no change in PL. When the ∆Pes/∆Paw ratio is within 0.8–1.2, the Pes measurement is considered to be reliable. Otherwise, the catheter should be repositioned and/or the balloon volume should be re-checked [3, 9]. In deeply sedated patients with or without paralysis, external manual compressions on the rib cage should be applied during an expiratory pause and simultaneous positive deflections of airway and esophageal pressures compared (positive pressure occlusion test) [1].

For the same esophageal catheter position, the Baydur and the positive pressure occlusion tests provide similar changes in airway and esophageal pressures [10]. The absolute value can be significantly higher when the esophageal balloon is placed in the lower third of the esophagus, compared to the middle one, due to the higher superimposed pressure generated by the heart and the lung. Moreover, the pressure generated by the esophageal wall as a reaction to balloon filling may increase the absolute value of Pes above the pleural pressure (Ppl) value. Therefore, a dedicated calibration procedure has been recently described that removes this artifact, resulting in the improved use of absolute values of Pes, especially when large-volume balloons are used [6]. While the presence of a nasogastric tube does not seem to significantly affect Pes measurement [11], nasogastric tubes equipped with an esophageal balloon are now available [4].

Transpulmonary pressure

Changes in transpulmonary pressure during mechanical ventilation: a measure of stress on the acute respiratory distress syndrome-injured lung

The clinical use of Pes measurement allows the estimation of PL as the difference between Paw and Ppl (estimated as Pes), which drives ventilation of the lung (including the airways). Here, we first look at the pathophysiological significance of the changes in PL induced by ventilation.

The airway driving pressure (the change between end-expiratory total pressure to end-inspiratory plateau pressure) delivered by the ventilator to inflate the respiratory system (∆Paw) may be divided in two components: one to insufflate the lung, called change in transpulmonary pressure (∆PL), and one to move the chest wall, called change in pleural pressure (∆Ppl), where ∆Ppl equals the changes of the esophageal pressure (∆Pes) [12, 13]:

$$ \Delta {\text{Paw}} = \Delta P_{\text{L}} + \Delta P_{\text{pl}} $$

In this section, we will refer to tidal ∆PL, ignoring for sake of simplicity the fraction of ∆PL due to positive end-expiratory pressure (PEEP). As the changes in lung volume and chest wall volume are deemed to be identical, the expression above may be written as:

$$ \Delta {\text{Paw}}/\Delta V = \Delta P_{\text{L}} /\Delta V + \Delta P_{\text{pl}} /\Delta V $$

The elastance of the respiratory system (Ers) is equal to the lung elastance (EL) plus the chest wall elastance (Ecw). Since Ers = ∆Paw/∆V and EL = ∆PL/∆V, the following expression may be derived:

$$ \Delta P_{\text{L}} = \Delta {\text{Paw }} \cdot \, (E_{\text{L}} /{\text{Ers}}) $$

This equation tells us that the pressure applied to the lung for a given driving pressure depends on the ratio between the EL to the Ers, irrespective of their absolute values. In normal subjects, the typical value of EL/Ers is equal to 0.5. In intensive care patients without respiratory or abdominal problems this ratio averages ≈0.7; in acute respiratory distress syndrome (ARDS) patients it may range from 0.5 (or even lower) to 0.9. Given these values, in a patient ventilated with PEEP 10 cmH2O and a plateau pressure of 30 cmH2O, the tidal ∆PL may range from 10 cmH2O {non-injurious, i.e. [(30 cmH2O plateau Paw −10 cmH2O PEEP) × 0.5]} to 18 cmH2O {highly “bio-traumatic”, i.e. [(30 cmH2O plateau Paw −10 cmH2O PEEP) × 0.9]}, as discussed further in this review [12]. Thus, the lung protective ventilation strategy might be, theoretically, titrated to a “safe” tidal ∆PL rather than to tidal volume [14].

When tidal ∆PL is applied, it changes the lung volume (∆volume). The ratio between ∆volume and the pre-inflation lung volume (V0) is called strain. The tissue molecules which compose the extracellular matrix develop a force opposing the tidal ∆PL, which is called stress. Stress and strain are nearly linearly related up to the total lung capacity. Therefore,

$$ {\text{stress}} = K \times {\text{strain}} $$

meaning that

\( \Delta P_{\text{L}} = K \times \Delta V/V_{0} \) [12, 15]

The proportionality constant K is called specific lung elastance and is the tidal ∆PL value at which V0 doubles its size (without PEEP). In humans K is around 12–13 cmH2O, which is substantially greater than other animal species [16, 17], and it seems to be similar in patients with ARDS and healthy subjects [15]. These concepts are summarized in ESM Fig. 1. As shown, when tidal ∆PL approximates 20 cmH2O, part of the lung enters its total lung capacity region. In healthy animals, when ventilation occurs in this area, lethal ventilator-induced lung injury develops within 24 h [17]. In clinical practice, it is rare to reach these values of tidal ∆P L. However, it must be remembered that in ARDS a substantial fraction of the lung includes inhomogeneities which act as stress raisers, with the possible effect of doubling the locally applied pressure [18]. Therefore, a tidal ∆PL of 10 cmH2O—in the presence of stress raisers—may regionally increase transpulmonary and transalveolar pressure to 20 cmH2O.

While waiting for more definitive evidence, tidal ∆PL should probably be kept below 15–20 cmH2O in patients with homogeneous lung parenchyma (normal, post-surgical patients) and possibly below 10–12 cmH2O in patients with inhomogeneous lung parenchyma (ARDS).

When respiratory muscle activity is present, then the active changes in Ppl can dominate the PL changes driving ventilation.

Transpulmonary pressure directly measured to guide mechanical ventilation in ARDS

Measurement of Pes also allows estimation of the absolute value of PL, measured as the Paw minus the Ppl and clinically estimated as Paw-Pes. The alternative elastance-derived method for estimating \( P_{\text{L}} (\Delta P_{\text{L}} = \Delta {\text{Paw}} \cdot \left( {E_{\text{L}} /{\text{Ers}}} \right) \) was covered in the previous section. In patients breathing on non-invasive ventilation, PL is estimated using the pressure in the mask (Pmask-Pes). Due to the location of the esophagus in the supine position, an increase in Pes caused by mediastinal weight of 3–7 cmH2O may exist in that position: whether Pes is higher than Ppl at the corresponding height in the chest is unclear [19]. Because Ppl varies spatially in the chest, Pes cannot theoretically represent Ppl over the entire lung. However, animal studies suggest that Pes represents well the pressure surrounding much of the lung [20].

In ARDS, ventilator settings are adjusted to reduce tidal overexpansion of the lungs by limiting tidal volume (and pressure) and to avoid repeated airspace collapse and derecruitment by maintaining adequate PEEP [21]. Pleural and intra-abdominal pressures vary widely and can be quite high in critical illness, regardless of chest wall compliance [22]. Therefore, a given PEEP applied by the ventilator could have varying effects on PL. A clinical trial in ARDS patients compared PEEP adjusted to maintain a positive end-expiratory PL (Fig. 1) with PEEP levels specified in the ARDSNet low-tidal volume table [23]. The PL approach significantly improved oxygenation and compliance and showed a trend toward reduced mortality [23]. This rationale was confirmed in animals with lung injury and different degrees of Pes elevation due to chest wall restriction. Targeting a positive end-expiratory PL in chest-restricted animals maintained lung volume and lung compliance, reduced hypoxemia and decreased pulmonary edema, mechanical abnormalities, pro-inflammatory mediator release and histological signs of ventilator-induced lung injury [24].

Fig. 1
figure 1

Directly measured transpulmonary pressure in acute respiratory distress syndrome (ARDS). Airway opening pressure (Pao), flow (Flow), esophageal pressure (Pes) and transpulmonary pressure (P L ) in a patient with ARDS on mechanical ventilation. At the end-expiratory occlusion (cursor) a total positive end expiratory pressure (PEEP) of 14 cmH2O and a Pes of 17 cmH2O result in a PL of −3 cmH2O. During the occlusion, a manual chest compression (PUSH) raises Pao and Pes equally, indicating adequate balloon placement in the chest

Uncertainties of direct esophageal manometry include the spatial variation in Ppl and the mediastinal weight that elevates Pes in the supine position [19]. As previously discussed, adequate filling and positioning of the balloon is important to obtain reliable measurements.

Transmural vascular pressure

Transmural pulmonary vascular pressures

The force distending vascular walls is called the transmural pressure (Ptm) and is determined by the difference between the pressure inside and outside of the vessel [25]. An increase in Ptm implies an increase in volume of the vessel. Since we are surrounded by atmospheric pressure (Patm), transducers are zeroed to Patm, and the pressure inside systemic vessels is the Ptm. For alveolar vessels, however, Ppl (not Patm) surrounds the vascular structures, and thus the value measured inside the vessel relative to Patm is no longer the Ptm.

Measurements of respiratory variations in Ppl obtained by esophageal devices can be used to determine changes in Ptm of pulmonary vascular structures that occur during the respiratory cycle [26, 27]. Measurements are usually obtained at end-expiration because this usually is when Ppl is closest to Patm. However, this may not be true when patients actively recruit expiratory muscles and increase Ppl during expiration (Fig. 2), or when Ppl is elevated by intrinsic or extrinsic PEEP. Failure to appreciate over-estimation of Ptm can lead to important management errors. For example, a central venous pressure of 20 mmHg and a low cardiac output suggest depressed cardiac function, and diuresis or use of an inotrope is likely to be appropriate. However, if Ppl is 20 mmHg and the Ptm is actually zero, the patient needs volume and not an inotrope or diuretic.

Fig. 2
figure 2

Value of esophageal pressure (Peso)-derived transmural vascular pressure (Ptm). Examples of changing Ptm of the central venous pressure (CVP) and pulmonary artery occlusion pressure (Ppao) during pressure support. Black horizontal bars mark inspiration. Paw airway pressure. The subject has marked active expiration on all breaths. a CVP is 10 mmHg at end-expirations (it is overlapping the Peso) but 6 mmHg at the start of expiration, which is the likely the Ptm. b 10 min later, CVP has increased to more than Peso, so that CVPtm (tm transmural) has increased, likely indicating recruitment of abdominal volume from the active expiration. Ppaotm did not change. c 1 min later, only Ppao is shown. The Ppaotm has decreased with some recovery late in expiration. It is likely that the rise in pleural pressure (Ppl) inhibited venous return with some later recovery because of the recruitment of abdominal volume. The subject likely was air stacking. Ptm measurements of CVP and Ppao would be very misleading without Peso. Double arrow indicates the change (∆) in Peso from the baseline value. Peso is measured with a fluid-filled catheter so that only the ∆Peso is used. All changes are relative to the values shown in a

Although transmural diastolic pressure of the right heart determines cardiac output, the directly measured value—and not Ptm—determines venous return. Thus, a central venous pressure elevated by PEEP reduces the gradient for venous return and cardiac output. Maintenance of cardiac output requires an increase in the upstream venous pressure by either a decrease in vascular capacitance or a fluid bolus [28]. However, there is a price to pay. The high venous pressure will increase capillary leak. Furthermore, when the patient is extubated and the positive Ppl is removed, the gradient for venous return increases and the lungs can be flooded if the left ventricle cannot handle the increased volume [29].

On the left side of the heart, increasing Ppl reduces the afterload on the left ventricle for the heart is effectively lifted relative to the systemic circulation [30]. This tends to increase cardiac output, but the benefit may be offset by a decrease in venous return due to the increase in Ppl. An inspiratory decrease in Ppl adds an afterload to the left ventricle for the heart that is effectively lowered relative to the rest of the body [31, 32]. This also increases venous return. If the swings are large enough or long enough, the rise in left ventricular filling pressure can produce pulmonary edema, especially when left ventricular function is poor.

In summary, changes in Ppl during the respiratory cycle need to be considered when interpreting hemodynamic pressures. Interpretations are thus greatly benefitted with a simultaneous measurement of Ppl.

Muscular pressure

Effect of spontaneous breathing on transpulmonary pressure during assisted ventilation

Spontaneous breathing can offer protective effects to the injured lung: improved oxygenation and increased dorsal aeration/ventilation [33]. This results from the increase in PL obtained by lowering Ppl (PL = Paw − Ppl) [32]. However, this protective role has generally been suggested in patients with less severe ARDS and modest spontaneous demands [34, 35]. Accumulating evidence indicates that spontaneous breathing might become deleterious in severe ARDS cases, as suggested by the beneficial effect of muscle paralysis observed in patients with severe ARDS [35, 36]. First, PL calculated on Pes is likely to be injuriously high in severe ARDS, combined with the already high Paw and uncontrollable spontaneous demands [35]. In contrast, in less severe ARDS, Paw can be maintained at relatively low values, and spontaneous demands are often modest, thereby preventing injurious increases in PL. Second, strenuous inspiratory efforts can shift air from non-dependent to dependent lung regions (i.e. pendelluft) even in the absence of large tidal volume and elevated Paw [37]. This occurs because the negative changes in Ppl generated by diaphragmatic contraction have localized effects in dependent regions, and the Ppl changes are not uniformly transmitted (i.e. solid-like behavior). To this end, lung recruitment might improve the correlation between Pes and the true maximal PL because it decreases atelectatic ‘solid-like’ regions.

Despite these limitations, monitoring Pes in ARDS patients with spontaneous inspiratory effort is highly relevant. First, PL calculated on Pes is a helpful measure for clinicians to detect the harm of spontaneous effort in severe ARDS, as suggested by large differences in PL between in mild and severe ARDS when spontaneous effort is preserved [35]. In this regard, even if no “safe” values for PL are reported, total end-inspiratory PL should probably be maintained below 20–25 cmH2O, which is known as the upper limit of physiological range [38]. Second, the intensity of pendelluft is proportional to the strength of the spontaneous effort (reflected by negative deflection of Pes) [37].

Finally, it is important to note that PL can be divided into two components: the resistive pressure, necessary to generate airflow between the airway and the alveoli, and transalveolar pressure, the pressure needed to expand the alveoli. Hence, only under static conditions (i.e. in the absence of flow) does PL equal the pressure distending the alveoli. Thus, unless mechanical properties change, PL should not differ between assisted and controlled ventilation for comparable tidal volumes and flows.

Pes measurement to estimate respiratory muscle effort

Contraction of the respiratory muscles generates a negative swing in Ppl, which can be assessed by changes in Pes (Fig. 3). The pressure generated by the respiratory muscles (Pmus) is computed as the difference between the static recoil pressure of the chest wall (Pcw,rel, obtained from Pes recordings under passive conditions) and the swing in Pes obtained under active conditions:

Fig. 3
figure 3

Respiratory effort indices derived from the esophageal pressure analysis. The figure schematically represents airway pressure (Paw), flow (V′) and Pes of a patient with intrinsic PEEP (PEEPi) under pressure support ventilation. The static recoil pressure of the relaxed chest wall (Pcw,rel), which represents the Pes that would be observed in passive conditions, has been superimposed to the Pes waveform. ❶ The difference between Pcw,rel and Pmus corresponds to the muscle pressure developed by the respiratory muscles (Pmus). ❷ Campbell diagram: the work of breathing (WOB) is computed as the integral of the Pmus over the volume displaced. The chest wall compliance (Ccw) represents the value of the esophageal pressure when the muscles are relaxed and the volume passively insufflated above the functional residual capacity (FRC). Dashed gray line crosses the pressure–volume waveform at the zero-flow points. The pressure developed at the onset of the inspiratory effort without generating any volume represents the PEEPi. These landmarks unveil the different components of the WOB: resistive (red area), elastic (blue area), related to the PEEPi (green area) and related to active expiration (grey area). ❸ The Pes pressure–time product (PTPes) corresponds to the area under the Pmus curve over the inspiratory time. The pressure developed between the start of the inspiratory effort and the onset of the inspiratory flow represents the amount of pressure needed to overcome the PEEPi. An additional pressure is dedicated to trigger the ventilator. The end of the active inspiratory muscle contraction occurred after the maximal negative value of the esophageal waveform and corresponds, in the case of ventilator cycling synchronization, to the onset of the expiration. In these conditions, relying the two points of zero-flow (dashed gray line) unveils the Pes related to the dynamic lung compliance. The PTPes can therefore be partitioned into the following parts: resistive (red area), elastic (blue area), related to the PEEPi (green area) and related to the inspiratory trigger (black area)

$$ {\text{Pmus}} = {\text{Pcw}},{\text{rel}} - {\text{Pes}} $$

The chest wall compliance (the passive volume/Pes relationship) can also be estimated from a theoretical value (4 % of the predicted vital capacity per cmH2O).

Values of Pmus between 5 and 10 cmH2O can be considered, as a crude estimate, to be the desirable respiratory effort during partial ventilatory support [39]. The work of breathing (WOB) of the respiratory muscles corresponds to the integral of the product of Pmus and the generated volume, represented graphically by the Campbell diagram [40] (Fig. 3):

$$ {\text{WOB }} = \int {{\text{Pmus}} \cdot {\text{d}}V} $$

WOB is expressed in joules and can be reported per liter or per minute. Normal values are around 0.35 and 2.4 J min−1, respectively [41]. WOB is computed based on volume displacement, and its value does not account for the duration of the respiratory effort and does not reflect the true energy expenditure of the respiratory muscles in the case of isometric effort.

The esophageal pressure–time product [42] (PTPes), which represents the integral of the Pmus curve over time, circumvents these limitations:

$$ {\text{PTPes }} = \int {{\text{Pmus}} \cdot {\text{d}}t} $$

PTPes is well correlated with the energy expenditure and oxygen consumption of the respiratory muscles. Normal values can be estimated to be between 50 and 150 cmH2O s min−1 [40]. Proper measurements of both WOB and PTPes take into account the presence of intrinsic PEEP [43].

Non-invasive estimates of respiratory effort may be useful at the bedside. In patients who are assisted with proportional assist ventilation mode, the Pmus value can be estimated from the value of the Paw and the gain set by the clinician [39].

Clinical applications

Clinicians may be interested in estimating respiratory muscle effort for several reasons. First, in the context of hemodynamic shock, respiratory muscle effort may account for a significant proportion of total body oxygen consumption, thereby limiting oxygen delivery to vital organs [44]. Ensuring adequate ventilation and sedation by monitoring respiratory muscle effort may help to optimize the distribution of oxygen delivery. Second, insufficient and (possibly) excessive levels of inspiratory effort during ventilation can injure the diaphragm [4549]. Titrating ventilation support to achieve near normal levels of inspiratory effort may prevent diaphragm injury and accelerate liberation from ventilation. Third, monitoring inspiratory effort may help clinicians to recognize potentially injurious patterns in patients with ARDS receiving partially assisted ventilation, both for the lungs and for the diaphragm [37]. The relevance of monitoring inspiratory effort during weaning trials is discussed in the section "Pes monitoring during weaning from mechanical ventilation" of this review.

Patient–ventilator asynchrony and the role of Pes monitoring

Mechanical ventilation, for the main part, is used as partial ventilatory support in situations where both the patient and the ventilator work together. Ideally, the ventilator settings should be adjusted to maintain a normal level of respiratory muscle activity and harmonious patient–ventilator interaction. Patient–ventilator asynchrony, which can be defined as a time mismatch between patient’s effort and the ventilator breath duration, often occurs during assisted modes of ventilation and but is unrecognized. Many studies have demonstrated that major asynchrony is a common occurrence (at least 25 % of the patients) throughout the course of ventilation [5052]. Thille et al. [52] demonstrated that asynchrony occurred during both assist-control and pressure support ventilation and that it was associated with a longer duration of ventilation in patients who had an asynchrony index indicating >10 % of completely asynchronous breaths [7.5 (interquartile range 3–20) vs. 25.5 (9.5–42.5) days]. An observational study in 50 patients by Blanch and colleagues [53] found that asynchrony was detected in all patients and in all ventilator modes. These authors reported that the mortality rates among patients in the Intensive Care Unit and hospital increased in patients who had high asynchrony index (>10 %); these patients also showed a trend toward longer duration of mechanical ventilation. Importantly, adjusting the ventilator settings can dramatically reduce asynchrony [54].

Detecting patient–ventilator asynchrony on the ventilator display without recording respiratory muscle activity is, however, not always easy and can be missed even by experts [55]. This difficulty justifies attempts to develop automated systems [56, 57] or to use an esophageal signal, such as the electrical activity of the diaphragm (EAdi) [58] or the Pes. The Pes recording can detect asynchrony by comparing the time occurrence of the changes in Pes with those of Paw and flow-time waveforms (Fig. 4). The major groups of asynchrony can be classified according to the phase of respiration as (1) trigger asynchrony, (2) cycling asynchrony and (3) flow delivery mismatch.

Fig. 4
figure 4

Esophageal pressure monitoring to diagnose patient–ventilator asynchrony. Tracings of airflow (Flow), Paw, Pes, and end-tidal carbon dioxide (EtCO 2 ) in patients receiving mechanical ventilation. a Delayed cycling: an increase in Paw and a sharp decrease in flow (arrows) were observed due to abrupt relaxation of inspiratory muscles at the end of inspiration (dashed vertical line). b Double triggering the ventilator delivered two consecutive triggered breaths, whereas only one inspiratory effort determined by Pes was observed. c Reverse triggering: the reverse triggered breath (dashed arrow line) occurred following the machine breath, demonstrating positive deflection of Pes followed by negative deflection of Pes (dashed vertical line). d Premature cycling: Pes revealed the mismatch between neural inspiratory time (Ti) and ventilator Ti (neural ≫ ventilator Ti), resulting in amputation of the peak expiratory flow and an increase in EtCO2 due to prolonged Ti (thick arrows). e Ineffective effort: patient’s inspiratory effort (small negative deflection of Pes) was observed but it was not sufficiently strong to trigger the ventilator; consequently, a distortion of airflow and Paw and an increase in EtCO2 were observed (arrowheads)

Trigger asynchrony

  1. 1.

    Ineffective effort is recognized by a negative Pes swing without ventilator pressurization. It occurs in situations combining auto PEEP and excessive ventilation, inducing low respiratory drive and low inspiratory muscle effort, and it can be aggravated by a poorly sensitive triggering system. It occurs frequently with pressure support ventilation and is associated with poor outcomes.

  2. 2.

    Double triggering is the occurrence of two consecutive triggered breaths separated by a very short expiratory time. It occurs when the neural inspiratory time (Ti) is longer than machine Ti, usually in patients with a high respiratory drive.

  3. 3.

    Reverse triggering is a phenomenon of respiratory entrainment and defined as a muscular effort triggered by the ventilator, usually in highly sedated patients. It occurs during the insufflation phase or at the transition phase from mechanical inspiration to expiration and can then lead to double cycling [59].

  4. 4.

    Auto-triggering is defined as a cycle delivered by the ventilator without triggering by the patient. This type of asynchrony can be seen in case of circuit leaks or cardiac oscillations.

Cycling asynchrony

  1. 1.

    Premature cycling is defined as a cycle with the machine Ti shorter than the neural Ti.

  2. 2.

    Delayed cycling occurs when the machine Ti is longer than the neural Ti

Flow delivery mismatch

  1. 1.

    Flow starvation occurs during volume control mode when the ventilator delivers a peak inspiratory flow that is lower than the patient’s demand. This is not an asynchrony per se, but it usually requires a change in the settings.

Patient–ventilator asynchrony is common and is associated with poor clinical outcomes. Pes can be used to better detect asynchrony at the bedside and to adjust the ventilator settings accordingly. However, whether the asynchrony is an indicator of severity or constitutes a biomarker for another systemic feature that compromises patients’ outcome warrants further investigation.

Pes monitoring during weaning from mechanical ventilation

The pathophysiology of weaning research has demonstrated that respiratory effort changes progressively as patients fail a weaning trial [60, 61]. During a trial of unassisted breathing, PTPes remained unchanged in patients who were successfully weaned from mechanical ventilation. In contrast, PTPes increased in patients who failed weaning because the mechanical load (resistive, elastic, and intrinsic-PEEP components) on the respiratory muscles increased [60, 62]. When the dominant functional abnormality was investigated in 17 patients who failed to be weaned, ten had worsening mechanics and muscle function, and four had blood gas abnormalities; failure in weaning could not be explained by functional abnormalities in the remaining three patients, suggesting that behavioral factors may have contributed to this failure [60].

The practice of medicine, though based on scientific knowledge, is centered on individualized care. The physician’s primary task is to figure out how scientific principles apply to one particular patient. By inserting an esophageal catheter in a difficult-to-wean patient, a clinician can determine whether worsening mechanics (most common cause) was responsible for weaning failure, and then take the necessary steps to correct the problem. If an increase in resistive work occurred during a failed trial, then the management plan would focus on decreasing airway resistance by using bronchodilators and frequent suctioning or diuretics if the increase in resistance was due to bronchial edema. An increase in the elastic load could indicate that subclinical cardiogenic pulmonary edema may have been responsible for the failed trial (ESM Fig. 2). Indeed, pathophysiological studies have shown that cardiac performance worsens during weaning failure as a result of an increase in left ventricular afterload, which, in turn, is due to more negative swings in Pes [29, 63, 64].

Jubran et al. [65] showed that throughout a failed weaning trial, swings in Pes showed greater changes than the rapid shallow breathing index (f/VT). Accordingly, monitoring of Pes throughout a weaning trial may provide additional guidance over a single measurement and, thus, may enhance the predication of weaning success/failure. When the change in Pes was quantified, the Pes trend index was more accurate in predicting weaning outcome than first-minute measurements of either Pes swings or f/VT [65]. If these results can be confirmed, the Pes trend index could provide a useful monitoring tool during weaning. Moreover, increases in Pes swings during the trial could alert a physician to institute therapy, such as bronchodilators, vasodilators or diuretics, at a time that might convert an unsuccessful trial into a successful one.

The correlation between diaphragm electrical activity and Pes swings

Obtaining reliable Pes measurements can sometimes be challenging, and the question of using other physiological signals to obtain similar information becomes relevant. The electrical activity of the diaphragm (EAdi), a direct expression of neural inspiratory activity, can be recorded in daily clinical practice by a nasogastric tube connected to a ventilator and equipped with electrodes [66]. EAdi could theoretically be used as an alternative to Pes to estimate inspiratory demand. Studies focusing on this issue have shown a tight correlation between EAdi and transdiaphragmatic pressure (Pdi) in healthy volunteers and obstructive patients at rest, even in the presence of increased inspiratory flow rates or lung volumes [6769]. EAdi and Pdi were also shown to be correlated and to be changed by the same magnitude during pressure support ventilation in a population of intubated patients suffering from acute respiratory failure when the level of assistance was increased [70]. The drops in both EAdi and Pes associated to external ventilatory assistance have also been correlated (Fig. 5) [71]. On the contrary, EAdi and Pdi amplitudes were found to be poorly correlated in patients with chronic obstructive pulmonary disease during incremental exercise [72], probably due to decreased neuro-muscular coupling in the presence of air trapping. In addition, it has been reported that the correlation factor (i.e. the slope of the linear correlation graph) between Pdi and EAdi differs between subjects, with one of the main reasons being individual anatomy [69]. Thus, EAdi amplitude can differ between subjects with similar inspiratory demand and requires calibration for inter-individual comparisons—for example, by dividing measured EAdi amplitude by the EAdi amplitude obtained during maximal voluntary inspiratory effort [69]. Two alternative calibration procedures that can be used for intubated patients with acute respiratory failure have recently been published [73, 74]. One of these, the Pmus/EAdi index (PEI), is measured during end-expiratory occlusion as the drop in Paw divided by the EAdi needed to generate it. By multiplying peak EAdi values by PEI, it is possible to calculate the absolute values of a patient’s inspiratory effort. The second procedure is the patient–ventilator breath contribution index [74], defined as the ratio between tidal volume and EAdi during an assisted breath divided by the same ratio measured during a subsequent unexpected non-assisted one. The correlation between the patient–ventilator breath contribution index and the ΔPes swings/ΔPTPes ratio is significant. These two calibration approaches to compare absolute values of a patient’s effort obtained from EAdi require further validation. It is also important to note that EAdi amplitude can be influenced by other factors, such as sedation [75] and clinical evolution [74].

Fig. 5
figure 5

Clinical validation of the correlation between electrical activity of the diaphragm (EAdi) and Pes swings at different levels of respiratory support. In two representative ARDS patients, amplitude of Pes swings and peak EAdi values increased by the same magnitude (right side of each panel) when: a the amount of CO2 removed by the extracorporeal membrane oxygenation (ECMO) system decreased by the reduction of sweep gas flow (GF), without change in pressure support; b the pressure support level delivered by the ventilator was reduced. PSV Pressure support ventilation

EAdi can also be used to accurately monitor patient–ventilator synchrony [55, 75, 76]. Finally, as the EAdi value recorded at the onset of the inspiratory flow (auto-EAdi) is tightly correlated with the patient’s auto-PEEP, as measured by Pes [77], EAdi can also be used to estimate auto-PEEP.

EAdi provides alternative and complementary information to Pes, but it cannot replace it. Thus, whether the practical advantages of using EAdi outweigh the additional information provided by Pes (e.g. PL) must be decided on an individualized basis.

Conclusions and perspectives

The multiple applications already described support the notion that Pes monitoring is an essential tool for personalized medicine and for improving the capability of physicians to diagnose complex clinical conditions. The results of recent studies suggest that bedside decisions regarding ventilator settings, lung protection, sedation, paralysis, weaning or even hemodynamic support might radically change if it were possible to consider the information coming from Pes monitoring. Some technical aspects of Pes measurement will need further refinements (e.g., those involving quantitative measurements of the chest wall properties).

Both the approach of using absolute Pes values and the dynamic approach in which only the swings in Pes are quantified have shown promising results for ventilation adjustments [12, 20]. While the direct absolute method is simpler and has been made more reliable through the use of newly developed catheters and accurate calibration procedures, the dynamic approach, by testing a volume change, might better unveil the pathophysiological characteristics of the respiratory system. However, both methods require a number of assumptions: for the absolute PL, that end-expiratory Pes is equal to the end-expiratory P pl and that negative values of end-expiratory absolute PL correspond to a tendency of the alveoli and/or small airways to collapse; for the dynamic approach (testing a volume change), that Ppl is equal to zero at zero PEEP and that the EL and Ers behave linearly throughout different lung volumes. In fact, especially when exploring the upper limits of lung stress, the two approaches have been found to produce conflicting results.

Synchronization with ventilator signals, automatic filling and automatic in vivo calibration are future developments that may enable this technology to be introduced into the daily clinical routine. Meanwhile, many of the clinical applications described in this review, with most of them tracking negative swings in Pes, such as the monitoring of asynchrony, over-assistance, strenuous efforts or difficult weaning, have been extensively validated, indicating that clinicians and patients likely benefit from the availability of this technology in daily clinical practice.