One of the main mechanisms proposed to explain RT is "entrainment", which is defined as coupling or alignment between the respiratory center of the patient and the ventilatory cycles programmed in the mechanical ventilator. Entrainment can be observed when there is stable phase coupling between the frequency of stimulation – in this case the passive insufflation ventilatory cycles – and the frequency of the autonomous oscillator (the respiratory rhythm of the patient), acquiring a periodic relationship with different patterns of appearance.

Respiratory entrainment appears to be produced mainly by stretching of the slow adaptation receptors and sustained activation of the Hering-Breuer reflex mediated by the vagus nerve.9 Animal experiments have shown that the abolition of the Hering-Breuer reflex by cooling or sectioning of the vagus nerve impedes this type of interaction.10 It is believed that passive insufflation triggers the Hering-Breuer reflex through stimulation of the stretch receptors in the upper airway, lungs and chest wall, providing afferent feedback to the respiratory center, which subsequently equals the frequency of the external stimulus.11 However, RT has also been described in bilateral lung transplant patients despite sectioning of the vagal afferent pathway — this suggesting that vagal feedback might suffice to activate RT, but is not the only route capable of doing so. The physiology of RT appears to be more complex than believed, and the possibility exists that one or more of these mechanisms may be implicated in a severely ill patient.

The entrainment pattern may be defined as the coupling relationship between the neural respirations and the ventilator cycles. In this regard, 1:1 and 1:2 patterns would be equivalent to a patient respiration in each ventilatory cycle and to a patient respiration every two cycles, respectively. Physiological studies and research in patients subjected to mechanical ventilation have shown the 1:1 relationship to be the most frequent and stable entrainment pattern, which is mainly identified in patients under deep sedation or in transition to awakening with more superficial sedation levels.5 On the other hand, less stable entrainment patterns such as 1:3, 1:4 and 2:3 are evoked with ventilatory patterns different from the intrinsic pattern of the respiratory center. These are more unstable and can be interrupted by irregular patterns. At present, little is known of the factors determining a given pattern or of how entrainment may vary within one same patient.

One of the methods for determining the stability of RT is the analysis of phase angle, which is simply a way to quantify the latency of response of the respiratory center. The phase angle can be calculated as:

The evidence of entrainment in lung transplant patients in which the vagal afferent pathways have been sectioned suggests that other mechanisms not mediated by the vagus nerve could play a role in the presence of RT.14

A study in brain death patients has shown that passive insufflations by the ventilator may generate delayed respiratory muscle activation, as occurs in the case of RT. In this context, the absence of respiratory impulses from the brainstem suggests that activation of the respiratory muscles after ventilator insufflation could be mediated by the stimulation of thoracic mechanoreceptors, passive chest movements and/or activation due to elongation of the respiratory muscles.15 Other mechanisms, such as spinal reflexes or the presence of a spinal respiratory pattern generator, potentially could also be implicated.11

Recent studies have shown that RT may exhibit different phenotypes.16 On analyzing the pressure/volume curve of the Campbell diagram in patients with acute respiratory distress syndrome (ARDS), Baedorf Kassis et al. identified four phenotypes according to the timing of activation and muscle relaxation within the respiratory cycle. The first phenotype is characterized by rapid activation and rapid relaxation (during inspiration or before reaching 50% of the exhalation phase). The second phenotype is characterized by RT with rapid activation in the inspiratory phase and slow relaxation (up to 50% of the exhalation phase) culminating in the expiratory phase. The most frequent phenotype in turn is characterized by rapid activation that produces the greatest contraction peak during the expiratory phase. Lastly, in the fourth phenotype, RT starts and ends during the expiratory phase. The effects of RT may vary in relation to its timing within the respiratory cycle. In this regard, the RT phenotypes that manifest during inspiration could cause increases in inspiratory transpulmonary pressure and volume, while RT occurring during expiration could increase expiratory transpulmonary pressure and incomplete exhalation.

On the other hand, and in contrast to the classical form of RT during controlled mechanical ventilation and deep sedation, some small case series have reported the presence of RT in spontaneous respiratory modes (e.g., pressure support and neurally adjusted ventilatory assist [NAVA]).17 In these cases, the presence of RT could occur due to auto-triggering and airway leakage,18 or due to relaxation of the expiratory muscles that induce contraction of the inspiratory muscles, simulating "pseudo-reverse triggering".19,20 These RT variants require confirmation through future research.

The visual inspection of ventilator curves is the most commonly used way to assess the interaction between the patient and the ventilator in clinical practice. However, the capacity of the different healthcare professionals to detect dyssynchrony from the ventilator curves is deficient and scantly feasible due to time restrictions, and the main problem is the poor reproducibility between evaluators.21,22 Nevertheless, several features allow us to identify the presence of RT and distinguish it from other spontaneous breathing efforts or types of dyssynchrony.

Independently of the ventilatory mode used, we need to establish how it starts the ventilatory cycle (whether started by the patient or by the ventilator). In order for RT to be considered as such, the cycle must be mandatory in the absence of signs of patient effort. The next step is to identify variations in pressure or flow, depending on the ventilatory mode. In volume control - continuous mandatory ventilation (VC-CMV) mode, we can observe variations in the pressure curve, such as a decrease in peak pressure, as well as loss of the expiratory flow peak with a rise in flow in the expiratory phase in the flow-time curve. If there is a programmed pause pressure, it may be altered by eliminating it or modifying its form (Fig. 2A). Likewise, in pressure control — continuous mandatory ventilation (PC-CMV) mode, we can identify variations in the flow-time curve; for example, a rise in flow during the inspiratory phase time interval in the decelerated flow, indicating a contraction of the diaphragm, which at the same time may or may not correspond to a mild drop in the pressure-time curve. In the flow-time curve we can observe an amputation of the peak expiratory flow and a rise in flow in the first third of the cycling phase (Fig. 2B). It is important to note that if contraction of the diaphragm is powerful enough and close to the end of the inspiratory cycle, it may give rise to a new trigger independently of the ventilation mode - a phenomenon known as double cycling caused by RT (Fig. 2C).

Figure 2.

Detection of reverse triggering by visual inspection. Panel A) Spontaneous ventilation in VC mode is observed, with the appearance of RT of middle activation and late relaxation, with 1:1 pattern; the red lines indicate the start of mandatory ventilation, and the broken blue lines indicate the response of the diaphragm. In the flow-time curve, the solid arrows indicate amputation of the expiratory flow; the star indicates attempted flow positivity during the expiratory phase. In the pressure-time curve, the broken arrows indicate the drop in peak pressure. Panel B) Ventilation in PC mode is observed, with the appearance of RT of middle activation and late relaxation, with 1:1 pattern; the red lines indicate the start of mandatory ventilation, and the broken blue lines indicate the response of the diaphragm. In the flow-time curve, the broken arrows show an increase in flow during deceleration; in the pressure-time curve, the solid arrows show a slight drop in airway pressure. Panel C) Ventilation in VC mode is observed, with the appearance of RT of middle activation and late relaxation, inducing double cycling; the red lines indicate the start of mandatory ventilation, and the broken blue lines indicate the response of the diaphragm. In the pressure-time curve, the star indicates the double cycling. Abbreviations. VC: volume control; RT: reverse triggering; PC: pressure control; Pes: esophageal pressure.

(0.25MB).

The reference standard for the identification of RT remains the use of the esophageal balloon (Peso), or some other method that identifies the timing of the activation of the inspiratory muscles, such as the electrical activity of the diaphragm (EAdi). These recordings are very useful, since they can detect dyssynchrony by comparing the time of occurrence of the changes in Peso or EAdi with the airway pressure and flow-time curves in each ventilatory cycle. Using Peso, we detect RT through the negative shift of the esophageal pressure following the start of the mandatory ventilatory cycle. Likewise, delayed activation in the EAdi curve could be interpreted as RT.23 Both types of catheters have been used to detect RT, though their utilization in clinical practice remains limited, since advanced training and technical expertise in acquiring the signals are required.5

Diaphragmatic ultrasound is an emergent noninvasive technique for the detection of asynchronies. Direct observation of the diaphragm through its muscle excursion allows us to identify the moment of contraction, which together with visualization of the mechanical ventilator curves can serve to identify RT.24,25 Although this is a promising technique, its use requires experience, and synchronization between the ventilator data and the ultrasound tracing is one of the main challenges.

The main interest in automatization of the detection of asynchronies is driven by their high incidence and the need for very prolonged observation periods. The methods described above usually provide only point or retrospective observations. In the case of RT, five automated methods have been evaluated to date and which use information from the ventilator curves,26 Peso27 and EAdi.9,28 The algorithms used are based on standards developed from the knowledge of asynchronies, and more recently on the use of computational neural networks.29 A summary of the different algorithms and models evaluated, as well as of their diagnostic accuracy in detecting RT, is provided in Table S1 (see electronic Supplementary material).

Different mechanisms could explain how RT may be harmful to the lungs. Firstly, superimposing spontaneous effort on ventilator respiration exerts an additive effect upon the transpulmonary pressure in charge of distending the lung parenchyma, and therefore on the resulting tidal volume (VT), mainly in pressure control modes where flow is variable. Baedorf Kassis et al. showed that RT can increase VT on average from 51 to 128 ml, and transpulmonary pressure between 3–7 cmH2O, depending on the phenotype when compared with passive respirations.16

Secondly, the occurrence of RT could be associated with double cycling and excessive tidal volume. During RT, the inspiratory effort of the patient may persist beyond the end of the inspiratory phase programmed by the ventilator. If this effort is sufficiently deep and prolonged, it may lead to double cycling, which is the occurrence of two consecutive inspirations without an expiration in between — resulting in increased VT.30 Observational studies have found that 35% of all double cycling events may be caused by RT, and that this could result in an increase of up to approximately 200 ml in extra VT.30–32

Thirdly, during respirations with RT, we may encounter a phenomenon known as pendelluft, characterized by an exchange of air from the non-dependent pulmonary zones to the dependent zones, causing abrupt tissue deformation and potentially excessive local stretching of the implicated dependent zone.33 Yoshida et al., using electrical impedance tomography in an animal model of lung injury, demonstrated that the pendelluft phenomenon can occur during RT. These investigators observed that air exchange from the non-dependent lung towards the more dependent zones can cause local overdistension up to two times greater than the distension observed in the same region during passive ventilation.34

Lastly, and in contrast to the potentially harmful impact, the presence of RT could have a beneficial effect on the maintenance of end-expiratory volume due to eccentric activation of the diaphragm. In an experimental model of acute respiratory failure, Pellegrini et al. studied the impact of diaphragmatic activation during expiration and found that expiratory diaphragmatic contraction reduces collapse and increases aeration of the lung in comparison with controlled mechanical ventilation without respiratory muscle activity.35 These results suggest that the diaphragm could exert a control effect upon expiration, helping to preserve lung volume at the end of expiration.

During RT, the inspiratory effort of the patient is activated by the ventilator, and in most instances, part of the patient effort takes place in the expiratory phase. During the expiratory phase, the inspiratory muscles (e.g., the diaphragm) elongate as the lung volume decreases. Thus, activation of the inspiratory muscles results in muscle activation during their elongation (eccentric contraction).36,37

The impact of these eccentric contractions of the diaphragm upon its function and structure could run in opposite directions. On one hand, RT could contribute to preventing disuse and atrophy of the diaphragm, since eccentric contraction is characterized by the generation of greater forces than other types of contractions in the face of a certain angular velocity. Furthermore, it has been shown that eccentric contractions require less motor unit activation and consume less oxygen for generating muscle force than concentric contractions.38,39 On the other hand, RT may prove harmful for the diaphragm when the contractions take place repeatedly during the expiratory phase concomitant to the decrease in lung volume and stretching of the inspiratory muscles. Studies in animal models have shown that intense muscle contraction during elongation is associated with a decrease in diaphragmatic contractility and the capacity of the diaphragm to generate pressure.40 Likewise, in an experimental model of acute lung injury, the occurrence of RT with high breathing effort was associated with significantly less diaphragmatic function and a high proportion of abnormal muscle fibers (e.g., muscle damage) in comparison with passive mechanical ventilation without breathing effort.41

The impact of RT on patient-centered outcomes is a current focus of study and research. Several observational studies have evidenced that RT may be associated with some positive outcomes.

Mellado-Artigas et al. studied the incidence of RT in 39 subjects during the first 24 h of mechanical ventilation, using the measurement of EAdi. Interestingly, on dividing the cohort of patients according to the median occurrence rate of RT, they found that those patients with a higher incidence of RT had better oxygenation and more probabilities of progressing towards assist mode ventilation or of being extubated in the next 24 h than patients with a low frequency of RT.9 On the other hand, Rodríguez et al., in an epidemiological study on patients with ARDS, found RT to be associated with a decrease in in-hospital mortality (hazard ratio 0.65; 95%CI: 0.57−0.73), thus suggesting that it may be a marker of improved outcomes in patients subjected to mechanical ventilation.42 At present, larger studies are being carried out to better understand the potential impact of RT and the clinical outcomes in critically ill patients (NCT 03447288).