Monitoring Asynchrony During Invasive Mechanical Ventilation

Insufficient Assistance: Patients With High Respiratory Drive

Flow Starvation

Spontaneous breathing during mechanical ventilation can be beneficial or deleterious, depending on the strength of the inspiratory effort and the severity of lung injury.^14-16 As patients recover, their need for ventilatory support varies: they demand a transition from more ventilator-controlled support to a support in which their spontaneous efforts trigger the ventilator, progressing to partial support and, eventually, to weaning; throughout this process, ventilation must be tailored to the individual to avoid excessive inspiratory efforts that can cause or aggravate lung or diaphragm injury (Fig. 1, Fig. 2, Fig. 3).^17-20 In patients with lung injury, gentle spontaneous effort improves gas exchange and aeration. However, vigorous inspiratory effort leading to huge increases in transpulmonary pressure can cause double-triggering and associated breath-stacking, resulting in lung injury due to increased lung stress and strain and to increased lung perfusion caused by augmented vascular distending pressures. Evidence, mostly from patients with ARDS, indicates that preventing injurious spontaneous efforts through neuromuscular blockade^21,22 or high PEEP²³ results in greater physiological levels of tidal volume (V_T) and transpulmonary pressure, respectively.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

Flow starvation seen on tracings of air flow, airway pressure (P_aw), and tidal volume (V_T) in a mechanically ventilated patient on volume-control continuous mandatory ventilation. Flow starvation (ie, inspiratory flow mismatching) occurs when the ventilator fails to meet the patient’s flow demand. In this case, ventilator air flow is set inappropriately low at 40 L/min. Inspiratory effort begins after a period of synchrony (line) during mechanical inflation manifested by a progressive decrease in P_aw, along with shortening of the expiratory time and breath-stacking (arrow). Notice a progressive return to synchrony (line).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

Flow starvation with double-triggering seen on tracings of air flow, airway pressure (P_aw), and tidal volume (V_T) in a mechanically ventilated patient on volume-control continuous mandatory ventilation. Flow starvation (ie, inspiratory flow mismatching) occurs when the ventilator fails to meet the patient’s flow demand. Inspiratory effort continues during mechanical insufflation (green arrow in the P_aw waveform) in each breath; sometimes it is strong enough to trigger a second mechanical insufflation (red arrow) without expiration (ie, double-triggering ) and consequent breath-staking (blue arrows). Notice that the V_T in breath-stacking is almost double with the associated elevated P_aw.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

Flow starvation seen on tracings of air flow, airway pressure (P_aw), and tidal volume (V_T) in a mechanically ventilated patient on volume-control continuous mandatory ventilation with decelerating flow. Flow starvation (ie, inspiratory flow mismatching) occurs when the ventilator fails to meet the patient’s flow demand. Inspiratory effort continues during mechanical insufflation in each breath. Notice a negative deflection in the P_aw waveform (arrow) with a different magnitude in every breath.

In patients with acute respiratory failure, even those receiving mechanical ventilation, altered gas exchange, high metabolic demands, or stimuli from mechanoreceptors and vagal inputs can cause high respiratory drive.⁵ In this clinical scenario, mechanical ventilation must be tailored to ensure appropriate inspiratory flow (usually 50–70 L/min), inspiratory time, and V_T. Inspiratory-flow mismatch occurs when the ventilator fails to meet the patient’s flow demand, usually because flow delivery is set inappropriately low or because the combination of V_T and inspiratory time fails to provide adequate air flow.^10,24,25 Inspiratory-flow mismatch is more common in modes where flow is unmodifiable, such as VC-CMV with constant flow.²⁶ Inspiratory-flow mismatch could be corrected by increasing ventilator flow delivery or by using the variable flow pressure-limited breath. In a seminal study, Ward et al²⁷ reported that substantial inspiratory muscle activity can occur during mechanical ventilation, particularly during assisted mechanical ventilation at low trigger sensitivity and flow. In this condition, inspiratory work of breathing accounted for nearly 65% of the total inspiratory work. Other authors have noted the importance of titrating the peak inspiratory flow to patients’ needs relative to respiratory center activity. Cinnella et al²⁸ compared the effects on the respiratory work rate of assisted ventilation, delivered either with a square or decelerating flow pattern in VC-CMV versus with a constant pressure in pressure-control CMV (PC-CMV). At high V_T, no difference between VC-CMV and PC-CMV was observed, suggesting that the effect of air flow is less relevant. However, at moderate V_T, lower peak flow in VC-CMV failed to meet patients’ demands, resulting in greater work of breathing than in PC-CMV. Interestingly, when moderate V_T and high inspiratory flow were applied, no significant differences were found between modes. MacIntyre et al²⁹ also reported that the variable flow pressure-limited breath was a better approach for matching vigorous patient efforts than adjusting a set flow on a conventional volume-cycled breath. Importantly, increasing inspiratory flow at similar V_T implies a decrease in inspiratory time.

Dyspnea or air hunger is the sensation of breathing discomfort.^30-33 Persistent inspiratory-flow mismatch can be associated with dyspnea, even when no respiratory signs are evident. Schmidt et al³⁴ reported that half of intubated or tracheotomized patients on mechanical ventilation for > 24 h experienced dyspnea, and dyspnea was associated with anxiety, continuous mandatory ventilation, and tachycardia. Increasing V_T or inspiratory flow improved dyspnea in 35% of subjects, and extubation failure was more common in those whose dyspnea persisted after adjustment of ventilator settings. Therefore, after deciding on an appropriate, safe effort for a given patient, dyspnea must be assessed. Trained physicians, respiratory therapists, and nurses, regardless of their competence and years working in the ICU, normally underestimate patients’ self-reported assessments.³⁵ Most patients cannot communicate, so bedside algorithms might be useful.³⁶ Finally, the potential consequences of low assistance include excessive respiratory muscle load and limbic, paralimbic, and cerebellar activation in the brain that can promote air hunger.^37-41

Short-Cycling: Double-Triggering and Breath-Stacking

Short-cycling is defined as a breath cycle where the inspiratory time is less than one half of the mean inspiratory time. Double-triggering (breath-stacking) refers to 2 complete inspirations separated by a very short expiratory time; however, the classic definition of double-triggering takes into account only the expiratory time (ie, lower than one half of the mean inspiratory time) without incorporating a V_T threshold.⁴² For this reason, some authors insist that double-triggering cannot be used as a synonym for breath-stacking, and that the term breath-stacking should be used only to refer to the clinical consequence of double-triggering when incomplete exhalation between breaths results in a higher-than-intended V_T.⁴³

The underlying mechanism in both short-cycling and double-triggering is a mismatch between the patient’s and ventilator’s inspiratory times that causes the patient’s inspiratory effort to persist beyond the end of the ventilator’s inspiratory cycle; thus, the inspiratory muscles are still active at the beginning of machine expiration, impeding the elastic recoil of the respiratory system from increasing alveolar pressure (Fig. 4). In this circumstance, peak expiratory flow is aborted, and, if the patient’s effort is sufficiently long and strong to reach the inspiratory trigger threshold, a second ventilator breath is delivered (Fig. 5). Consequently, to a similar pathophysiological mechanism, which of both asynchronies develops depends mostly on the ventilator mode and setting. For instance, when a sustained inspiratory effort persists or is strong enough to overcome the inspiratory trigger threshold in PSV, short-cycling or double-triggering could develop, because in PSV a decrease in inspiratory flow from its peak to a preset threshold triggers the ventilator’s expiratory cycle. By contrast, in VC-CMV or PC-CMV, once the trigger threshold is reached, a complete second breath will be delivered and double-triggering with breath-stacking will occur, unless a preset pressure or volume alarm avert the second mechanical breath.^44,45

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

Short or premature cycling seen on tracings of air flow, airway pressure (P_aw), and esophageal pressure (P_es) in a mechanically ventilated patient on pressure-control continuous mandatory ventilation. Every breath delivered is patient-triggered (ie, assist ventilation) with a decrease in flow, P_aw, and P_es. Short or premature cycling develops when the neural inspiratory time is greater than the mechanical inspiration time. In this example, the inspiratory effort (solid line) continues after the mechanical insufflation ended (dashed line), with the negative deflection of the P_es persisting after the mechanical inflation terminates. This pattern develops in the 4 illustrative breaths. Image courtesy of Tai Pham and Laurent Brochard, Toronto, Canada.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 5.

Relationship between tidal volume (V_T) and peak inspiratory pressure (PIP) in a normal breath and in a double-tirggered breath, with or without reverse-triggering. Descriptive notched boxplots for V_T (top) and PIP (bottom) in each ventilatory mode. (A, C) Breaths without double-triggering (DT) versus breaths with DT breaths. (B, D) Breaths with reverse-trigger–induced DT breaths (DT RT) versus patient-triggered DT breaths (DT PT). Dots represent means, and box plots indicate medians and 25th–75th percentiles. PCV = pressure-control continuous mandatory ventilation; VCV = volume-control continuous mandatory ventilation with constant flow; VCVDF = volume-control continuous mandatory ventilation with decelerated flow. Notice the higher V_T, in both patient-triggered and reverse-triggered ventilation, delivered in double-triggered breaths without meaningful changes in PIP. From Reference 44, with permission.

Although uncommon, double-triggering is the second most frequent asynchrony.^4,42,44 Longer inspiratory time and higher inspiratory peak flow are associated with less double-triggering, probably suggesting better matching among neural inspiratory time, ventilator inspiratory time, and inspiratory demand. Higher PEEP is also associated with less double-triggering, probably because higher lung volumes decrease inspiratory efforts. On the other hand, low peak pressure is associated with more double-triggering, secondary to flow asynchrony.^44,46

Activation of inspiratory muscles during mechanical deflation results in an eccentric contraction of the diaphragm that can injure respiratory muscles.^6,47-49 When breath-stacking occurs, the higher V_T delivered can generate high transpulmonary and transvascular pressure gradients, as well as elevated local stress and uneven distribution of pressure in dependent zones of the lung (Fig. 3),¹⁴ which could cause ventilator-induced lung injury^14,44,46,50 Double-triggering can occur in clusters in patients with ARDS⁴³ or in general ICU patients,^44,46 but the prevalence of short-cycling or double-triggering or reverse-triggering is unknown.

Overassistance: Patients With Low Respiratory Drive

Ineffective Efforts

Ineffective efforts (IEs) are the most frequent type of asynchrony, affecting nearly 50% of mechanically ventilated patients.^11,42,51,52 Defined as an inspiratory muscle effort that is not followed by a ventilator breath, IE occurs when the patient’s attempt to initiate a breath does not reach the ventilator’s trigger threshold (ie, when the ventilator fails to detect the patient’s inspiratory effort). Patients experiencing IE often seem calm, without signs of dyspnea. For diagnostic purposes, a decrease in airway pressure associated with an increase in flow during expiration on tracings is highly suggestive of IE; a negative deflection in esophageal pressure or an increase in the electrical activity of the diaphragm not followed by a mechanical breath is confirmatory.

IE can result from different clinical conditions and reflect different situations. The main causes are delayed cycling, overassistance, and hyperinflation, all of which lead to intrinsic PEEP.^52-55 Decreasing respiratory drive by other mechanisms such as sedation increases the occurrence of IE, and deeper sedation with propofol is associated with higher frequency of IE during PSV (see Management).^56,57 Entrainment and reverse-triggering could also induce IE if they occur in the late inspiratory phase or early expiratory phase (see Reverse-Triggering: Entrainment Phenomenon).

The main clinical consequence of IE is muscle damage. In the context of overassistance, IE can lead to myofibrillar atrophy and contractile dysfunctions in a short time (overassistance myotrauma).^15,47 Excessive contractile loading while the muscle is lengthening (eccentric contractions) are particularly injurious; this eccentric myotrauma occurs when the muscle contracts actively during the ventilator expiratory phase.^47,54,58 The higher levels of support and increased V_T provided by PSV decrease dyspnea, resulting in decreased drive and a progressive increase in IE.⁵³ Plotting the level of assistance in PSV and dyspnea yields a U-shaped curve (Fig. 6); if assistance is too high or too low, discomfort increases.^31,51 Thus, clinicians should individualize support, aiming for the midpoint that avoids both excessive and insufficient assistance.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 6.

Relationship between comfort and level of assistance. Patient comfort was assessed with 2 different methods: (A) Borg scale; (B) visual analog scale. Notice that the extreme levels of assistance during pressure support ventilation generated the least comfort. From Reference 51, with permission.

Delayed or Prolonged Cycling

Prolonged cycling occurs when mechanical insufflation continues after the end of neural inspiration, sometimes even when the patient is actively exhaling (Fig. 7). When this occurs, the patient may actually fight the ventilator, recruiting expiratory muscles in an attempt to force expiration. Prolonged cycling is associated with the failure of the subsequent inspiratory effort to trigger the ventilator.⁵⁴

• Download figure
• Open in new tab
• Download powerpoint

Fig. 7.

Flow, airway pressure (P_aw), and inspiratory and expiratory muscle activity in a patient with COPD who received pressure support ventilation with delayed or prolonged cycling. The electromyograms in the lower portion of the figure show inspiratory muscle activity in the patient’s diaphragm and expiratory muscle activity in the transversus abdominis. The patient’s increased inspiratory effort caused P_aw to fall below the set trigger threshold, and inadequate delivery of flow by the ventilator resulted in a scooped contour on the P_aw curve during inspiration. Although the ventilator was still pumping gas into the patient, his/her expiratory muscles were recruited, causing a bump in the P_aw curve. Notice that neural inspiratory time is not coupled with machine inspiratory time; after the neural inspiration finished, the ventilator still provided air flow while the patient’s expiration begins. The dotted line shows P_aw in another patient who generated just enough effort to trigger the ventilator and in whom there was adequate delivery of gas by the ventilator. From Reference 3, with permission.

In PSV, the ventilator expiration cycle begins when flow decreases to a set percentage of peak inspiratory flow. Insufflation tends to be longer with higher levels of support and with increased air flow resistance. Additionally, higher levels of support result in a higher peak flow that may shorten the neural inspiration time, exacerbating the mismatch between the patient and the ventilator.⁵⁹ In this scenario, the inspiratory flow delivered decreases very slowly, and the flow-cycling mechanism can cause the ventilator’s inspiratory time to be longer than the neural inspiratory time. This overassistance results in hyperventilation, hypocapnia, and respiratory alkalosis, reducing respiratory drive and increasing asynchronies in a vicious circle.⁶⁰ Furthermore, the continuous recruitment of expiratory muscles is associated with failed weaning trials.^61,62 Finally, other, nonrespiratory mechanisms, such as airway leaks, can also cause prolonged cycling, so it is important to check endotracheal tubes and cuffs, tracheostomy tubes, and the entire artificial circuit.

Reverse-Triggering: Entrainment Phenomenon

The respiratory center is a biological oscillator, so it is susceptible to entrainment from the periodic imposition of input from the ventilator; in other words, periodic lung inflation can establish a fixed repetitive temporal relationship between the neural and mechanical respiratory cycles.⁶³ This phenomenon has been observed in rabbits,^64,65 cats,^63,66 dogs, and humans.^67,68 Entrain-ment develops more often when the ventilator’s V_T, breathing frequency, and inspiratory flow are very similar to the subjects’ neural breathing pattern. The pathophysiology of entrainment probably involves the slowly adapting and rapidly adapting stretch receptors in the lungs as well as vagal C-fibers. Nevertheless, entrainment has also been observed in vagotomized lung transplant recipients⁶⁸ and in brain-dead patients in the absence of respiratory drive from the brainstem,⁶⁹ so respiratory muscle activation after ventilator insufflation is likely mediated by various afferents.

Recently, the term reverse-triggering was coined to describe an asynchrony that results from respiratory entrainment.⁷⁰ In reverse-triggering, mechanical insufflation elicits a neural response, resulting in neural and muscular activity that ends in a ventilator-induced diaphragmatic contraction. It is important to emphasize that entrainment is not an asynchrony, but rather the mechanism underlying reverse-triggering. Moreover, once entrainment develops, reverse-triggering could induce other asynchronies, such as IE and double-triggering, so from a clinical viewpoint it could be useful to distinguish IE, double-triggering, and breath-stacking induced by invasive ventilation from those developing through other pathophysiological mechanisms (Fig. 8).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 8.

Reverse-triggering and elimination of reverse-triggered breaths during an end-expiratory hold: airway pressure (P_aw), flow, esophageal pressure (P_es), and transpulmonary pressure during 6 breaths and an expiratory hold. The first 4 breaths show signs of respiratory entrainment; there seems to be a 2:1 pattern of mandatory breaths to reverse-triggered breaths that start as a passive insufflation but are rapidly followed by patient effort, as can be concluded from the drop in P_es (ie, the mandatory insufflations elicit a neural response that leads to patient effort). Larger drops in P_es than in P_aw trigger the ventilator before exhalation of the previous breath, leading to breath-stacking with high tidal volumes. These breaths with patient effort are likely reverse-triggered. An expiratory hold was administered at the first asterisk and continued until the next asterisk, which abolished respiratory muscle activity for > 12 s. The arrow on the transpulmonary pressure (P_L) curve marks the moment that inspiratory effort was expected if patient effort was generated spontaneously. After the expiratory hold, another mandatory insufflation was administered by the ventilator; patient effort immediately follows. This is suggestive of reverse-triggering because the patient effort seems to depend on the external stimulus provided by mandatory ventilator insufflation. From Reference 71, with permission.

Reverse-triggering can be classified by the ratio of mechanical breaths to diaphragmatic contractions. The most common pattern observed in animal models and in humans is 1:1, but ratios of 2:1 and 3:1 have been reported.^64,66,70 The only reliable way to recognize reverse-triggering is by detecting the muscle contraction after mechanical insufflation in esophageal pressure measurements, although careful analysis of ventilator waveforms could also indicate reverse-triggering.^44,70-72 Automated algorithms promise to detect reverse-triggering from ventilator waveforms reliably and quickly without the need for esophageal pressure monitoring.⁷³

Reverse-triggering can lead to lung and muscle injury. Once IE or double-triggering induced by invasive ventilation develops, the pathophysiological mechanism of injury is the same as when these asynchronies develop from other causes. Su et al⁷⁴ reported that 30% of subjects with ARDS exhibit reverse-triggering that developed in the late inspiratory phase (41%) and early expiratory phase (59%), and that reverse-triggering was associated with breath-stacking with a larger proportion of V_T and maximum transpulmonary pressure fluctuations compared to passive breaths, suggesting that the mechanism of injury is not only volutrauma but also barotrauma. Interestingly, in one subject with ARDS, Yoshida et al⁷⁵ observed that reverse-triggering without breath-stacking elicited an injurious inflation pattern with asymmetric stretch of the dependent lung at constant V_T.

At present, our understanding of the prevalence and clinical meaning of reverse-triggering in mechanical ventilation patients is very limited because this phenomenon was discovered only recently, so few clinicians are familiar with it and trained to identify it.^70,76 Observational studies have often reported reverse-triggering in heavily sedated subjects with ARDS, but less information is available about the prevalence of reverse-triggering in other clinical scenarios.^70,74,75 Although de Haro et al⁴⁴ reported that double-triggering induced by invasive ventilation occurred in 34.6% of a cohort of 67 general ICU subjects, further studies are needed to investigate this issue in different groups of patients to better understand reverse-triggering and its clinical implications.

Assessment of Asynchronies

Asynchronies can be detected with different techniques, including the analysis of flow-time, pressure-time, and integrated-volume waveforms from ventilator screens; tracings from balloon-tipped esophageal catheters; electrical activity of the diaphragm, which is available on some ventilators; and more recently, automated algorithms (10).

The sensitivity and positive predictive value of analyzing breath-to-breath waveforms are very low (22% and 32%, respectively); even trained physicians recognize less than one third of asynchronies. Clinicians’ ability to identify asynchronies by visual inspection is partially influenced by their expertise and the type of asynchrony.⁷⁷ Adding analyses of esophageal pressure tracings could improve the recognition of IE, short-cycling, reverse-triggering, and high efforts during flow asynchrony.^78,79 Unfortunately, al-though it is minimally invasive, esophageal pressure monitoring is not routine in mechanically ventilated patients, and proper balloon placement and data interpretation require experience. The electrical activity of the diaphragm provides reliable information about patients’ inspiratory and expiratory time, but it is only available in Servo (Maquet, Sweden) ventilators in the neurally adjusted ventilation assist (NAVA) mode, and the position of the catheter must be checked periodically for displacement and adjusted to ensure correct positioning.⁸⁰ More importantly, it is not only the recognition of asynchronies that matters, but also their frequency and distribution over time; asynchronies can occur anytime during mechanical ventilation and often occur in clusters alternated with periods without dyssynchronies (Fig. 9).^{4,43,44,81,82}

• Download figure
• Open in new tab
• Download powerpoint

Fig. 9.

Distribution of asynchronies during mechanical ventilation. The asynchrony index, noted as percentage (%) per hour, was recorded continuously over several days in 4 representative patients. Recordings show that periods of very low asynchrony alternated with periods of high levels of asynchrony. From Reference 4, with permission.

Various automated algorithms have been developed and validated to detect asynchronies through the continuous analysis of flow-time and pressure-time tracings, although their performance varies.^{4,44,73,83,84} Another algorithm is based on the automated spectral analysis of flow signal.⁸⁵ Given the clustering of patient–ventilator asynchronies, automated detection, prediction, and intelligent alarms promise to be more accurate than time-consuming visual inspection of waveforms, which can only detect dyssynchronous events occurring during the actual observation period.

Asynchronies and Sleep in Critically Ill Patients

Nearly all patients in medical, cardiac, and surgical ICUs have fragmented sleep, with long sleep-onset latency, rapid-eye-movement latency, and poor sleep efficiencies.^86-90 Although the amount of sleep over a 24-h period may be relatively normal (7–9 h), short bouts during the day account for approximately 50% of this total.^91,92 Deranged sleep architecture is associated with ICU delirium.⁹³ In mechanically ventilated patients, noise, stress related to increased difficulty in communicating, dyssynchronous breathing, ventilator modes, drugs, and discomfort can further degrade the quality of sleep.^91,94,95 Meza et al⁹⁶ and Parthasarathy et al⁹⁷ reported that sleep fragmentation could be induced by increasing support on PSV. This was explained by a decreased drive during sleep but with the same support that resulted in overassist, thus causing hypocapnia. Thille et al⁹⁸ recently reported that time to weaning was 3 d longer in patients with atypical sleep.

Hypothesizing that asynchronies cause sleep fragmentation, 3 studies compared newer, proportional modes of mechanical ventilation designed to im-prove patient–ventilator synchrony (PAV+^99,100 and NAVA¹⁰¹) with PSV. All 3 studies reported that the newer modes reduced asynchronies during sleep, but the effects on sleep quality were small, improving it in 2 studies^100,101 and worsening it in the other.⁹⁹ However, as Younes⁹⁴ pointed out, fighting the ventilator during wakefulness may prevent patients from falling asleep until they are no longer able to fight, are sedated, or are exhausted. Hence, the deleterious effect of an unsuitable ventilation mode may be present only when the patient is awake, and the impact would be reduced sleep time and not poor-quality sleep. Our understanding of the relationships between patient–ventilator interactions and sleep is currently growing and constantly evolving.

Management

As is common in clinical medicine, better understanding of the mechanisms underlying asynchronies can improve treatment. Respiratory drive in ICU patients can be affected by many factors, including pain, temperature, inflammatory chemokines, deranged gas exchange, metabolic demands, neurological insults, and even behavioral factors and sleep state, thus modifying mechanical feedback (ie, the effects of volume and flow delivered by the ventilator upon the respiratory muscles), reflex feedback (ie, the influence of various receptors located in the lungs, chest wall, and respiratory tract upon delivered volume and flow), chemical feedback (ie, the ventilatory response to chemical stimuli driven by CO₂, pH, and O₂), and behavioral feedback (ie, the effects of cortical output on control of breathing).^{1,2,13,40,102-104} Thus, before attempting to correct an asynchrony, clinicians should address modifiable medical issues with careful examination of ventilator performance and the circuit to allow optimal neuromechanical coupling whenever possible.

The 2 most common approaches to dealing with asynchronies are sedation or analgesia and adjusting ventilator settings. Clinicians list promoting patient–ventilator synchrony among the main reasons for administering sedatives, but increasing sedation is associated with worse short- and long-term outcomes.^105-108 Thus, a more conservative strategy minimizing the use of sedatives during mechanical ventilation is currently recommended.^109-112 Sedatives influence respiratory center output by affecting either the respiratory drive or timing. When deeper levels of sedation diminish the respiratory drive, IE develops.^56,57 One study reported that the IE index increased by 2.7% for every 1-unit decrease in the Richmond Agitation-Sedation Scale.⁵⁷ Another report noted that the extent to which propofol reduced neural drive varied with the rate of infusion and that deeper sedation increased IE in PSV but not in NAVA.⁵⁶ The results of a recent study considering the entire period of mechanical ventilation in 79 subjects, in whom > 14 million breaths were analyzed,¹¹³ shed new light on the usefulness of sedatives and opioids in treating asynchronies (Fig. 10). Sedatives, whether administered alone or together with opioids, did not decrease asynchronies beyond what was achieved with opioids alone. Moreover, opioids were inversely associated with IEs, double-triggering, and asynchrony index, but when sedatives were added to the regimen, increased doses of sedatives were associated with higher IE and asynchrony index values. Additionally, unlike opioids alone, sedatives, used alone or in combination with opioids, were inversely associated with the level of consciousness. Thus, increasing sedation to improve patient–ventilator interaction might have the opposite result. The different effects of sedatives and opioids on asynchronies, level of consciousness, and dyspnea are presented in Table 1.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 10.

Relationship between asynchrony and dose of sedatives (left) and opioids (right). On opioids-only days (white circles in B, D, F), the opioid dose was inversely associated with AI, IEE, and DC. On sedatives + opioids days (black circles in A, C, E), opioid dose was still inversely associated with AI, IEE, and DT. Sedative + opioid dose was directly associated with AI and IEE (black circles in A, C), but not with DT (black circles in E). On sedatives-only days (white cirlces in A, C, E), the sedative dose was only inversely related to DT. AI = asychrony index; IEE = ineffective expiratory effort; DT = double-triggering. From Reference 113, with permission.

Dealing with the underlying cause of asynchronies by adjusting ventilator settings should be the preferred approach. A single-center study in patients receiving VC-CMV, most with breath-stacking, reported that ventilator adjustments were more effective than increasing sedation.⁵⁰ Table 2 summarizes the different ventilatory amendments that could be used to correct each type of asynchrony. It is important to consider the physiological concepts involved, because ventilator output affects patient output and vice versa, so any change in ventilator settings (eg, V_T, flow, inspiratory time, rise time, etc.) will alter the activity of the mechanoreceptors and neuromechanical feedback.

Managing flow asynchrony could be particularly challenging. It might be tempting to correct flow starvation by increasing the flow delivered; however, in parallel experiments, Laghi et al¹¹⁴ and Fernandez et al⁵⁹ reported that increasing the flow at a constant V_T decreases neural inspiratory time, thereby increasing the breathing frequency. Thus, managing high IE due to insufficient flow requires not only increasing the delivered flow to mitigate flow starvation, but also close observation of the resultant breathing pattern to avoid dynamic hyperinflation due to insufficient expiratory time.^45,52 Moreover, inspiratory-flow mismatch could be related to inadequate pressurization of the system unrelated to the patient’s inspiratory flow demand.

If a patient can breathe spontaneously, proportional modes might offer some advantages in certain patients. Several studies reported fewer major asynchronies and a greater proportion of time in synchrony with the ventilator with PAV+ than with PSV, even in patients with COPD.^99,115-117 Numerous studies found similar results for NAVA versus PSV.^118-121 Additionally, an ongoing study to determine whether PAV+ could improve outcomes is currently enrolling subjects (ie, the PROMIZING study; ClinicalTrials.gov: NCT02447692).

Another promising approach is partial neuromuscular blockade. In a proof-of-concept study of subjects recovering from ARDS who were able to breathe spontaneously in PSV but developed high V_T (> 8 mL/kg predicted body weight), continuously infusing low doses of rocuronium facilitated lung-protective ventilation by decreasing V_T and transpulmonary pressure while reducing work of breathing and maintaining diaphragmatic activity.²² Even though asynchronies were not specifically measured, the physiological basis of this approach (ie, diminishing vigorous diaphragmatic contractions induced by high respiratory drive) suggests that when high drive cannot be controlled, regulating output (ie, high effort) could be effective; nevertheless, future studies should investigate this possibility.

Outcomes

Although many studies have explored possible associations between asynchronies and outcomes, none have been specifically designed to assess the impact of asynchronies on outcome. One study reported that the duration of mechanical ventilation was longer in subjects with an IE index > 10%, although these conclusions were based on recordings of only 10 min during the first day of mechanical ventilation.¹²² Another study, based on 30-min recordings, also reported that the duration of mechanical ventilation was longer in subjects with asynchrony index > 10% who also had a higher incidence of tracheostomy, although mortality was unaffected.⁴² By contrast, in a prospective noninterventional observational study using dedicated software to analyze asynchronies during mechanical ventilation in 50 subjects, Blanch et al⁴ reported that subjects with asynchrony index > 10% had rates of re-intubation and tracheostomy similar to subjects with fewer asynchronies, but higher ICU and hospital mortality and a trend toward longer duration of mechanical ventilation. Although this study used extensive real-world data, its design precludes definitive conclusions, and dyssynchronies could have been related to disease severity rather than a direct cause of mortality.

Asynchronies tend to occur in clusters between relatively uneventful periods, and clustering might be more important than overall rates for outcomes. Periods with almost no asychrony alternate with high rates of asynchronies around the clock, during the entire duration of mechanical ventilation.⁴ Other studies have reported clusters of particular asynchronies (eg, double-triggering and IE). Beitler et al⁴³ noted that both the overall and peak hourly frequency of breath-stacking differed considerably among subjects. Defining clusters of double-triggering as ≥ 10% double-triggered breaths within a 3-min period, de Haro et al⁴⁴ reported that 59.7% of subjects had clusters, with a median of 6 cluster events per subjects and 41 double-triggered breaths per cluster. Defining clusters of IE as > 30 IEs in a 3-min period, Vaporidi et al⁸¹ observed that clusters were associated with increased risk of remaining on mechanical ventilation for > 8 d after the first recording and with increased risk of dying in the hospital. Moreover, other cluster characteristics, such as power and duration, also correlated with the duration of mechanical ventilation after the first recording, suggesting that the intensity of the clusters in a given time frame might be more important for outcomes than the general presence of asynchronies. In line with this hypothesis, using Bayesian joint models to assess the association between repeatedly measured variables (ie, SOFA and asynchrony index) and time-to-event data in a cohort of 139 mechanically ventilated subjects, Rué et al¹²³ reported a strong relationship between SOFA and vital status, but there was not enough evidence to conclude that asynchronies provide a more accurate indication of death prognosis than SOFA score alone (Fig. 11).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 11.

Analysis of the mixed effects of SOFA score and asynchrony index in ICU survival at ICU using a Bayesian joint model of bivariate longitudinal and competing risks data. SOFA score is strongly related with survival status at ICU discharge. The distributions do not overlap, indicating strong association of the SOFA score with respect to vital status. In contrast, asynchrony index at time t does not seem to be associated with vital status as clearly when overall severity is taken into account. For the event “Survived,” the asynchrony index has a positive support, indicating that patient–ventilator interactions could be a sign of good prognosis. For the event “Died,” the asynchrony index parameter indicates that patient–ventilator interactions at time t do not provide a more accurate indication of death prognosis than the SOFA score alone. From Reference 123, with permission.

Moreover, it is likely that asynchronies occurring early during mechanical ventilation when patients are ventilated with control modes have a different impact than those that occur during partial ventilatory support or during periods of liberation from mechanical ventilation. Therefore, to better understand the influence of different asynchronies on ICU and hospital outcomes, further studies should assess how they present (ie, clusters, intensity) and when they occur.

Future Directions

Recent definitions of personalized medicine consider physiological variables to be essential biomarkers in critical care because they are closely related to physiological recovery and organ support.^124,125 Precision critical care medicine begins with real-time evaluation of continuously recorded physiological variables, which provides valuable information about the patient’s current and dynamic risk profiles.¹⁰

Smart systems capable of automatically recording and evaluating physiological data in real time to improve the quality, safety, and efficacy of medical decisions are of great interest.^126,127 These systems must ensure that all relevant data are recorded and that information is intelligently filtered and presented appropriately.¹²⁷ To enable individualized assessment using physiological biomarkers to predict whether a particular patient is likely to experience an adverse event, systems must be capable of automatically transforming raw data into useful information and presenting it in integrated displays showing the information that is most relevant for clinical decision making.^124,127-129 Alert thresholds derived from continuous predictive analytical monitoring can be operationalized as a degree of change from the patient’s own baseline rather than from arbitrary cutoffs.¹³⁰

To develop smart alarms, artificial intelligence can incorporate multiple datasets, including demographics, mechanical ventilation data, and medical texts (structured or unstructured). Smart alarms derived from monitoring asynchronies during mechanical ventilation promise to facilitate decision making, optimize patient–ventilator interaction, decrease morbidity and mortality, and improve quality of life.^10,19,107

Exploiting the vast quantities of data generated in health care will change medical care, enriching physician–patient relationships with insights from machine learning.¹³¹ Furthermore, time-series forecasting aims to predict future events on the basis of past observations. The Hidden Markov model is a statistical ap-proach applied in data analysis to recognize patterns over time, and sequential data could be presented as a Markov chain of latent (or hidden) states. Recently, Marchuk et al¹³² reported the feasibility of Hidden Markov models to predict periods with high frequencies of asynchrony events. Importantly, this model could be used to create an alarm system that could predict the likelihood of asynchrony events and alert care providers. In addition to big data techniques, new approaches such as deep machine learning, artificial neural networks, and entropy and complexity analysis techniques will help clinicians better select patients who might benefit from disconnection from mechanical ventilation.

Recently, Tobin wrote, “Over the breadth of my pulmonary and critical care practice, no area demands greater understanding of physiological principles than ventilator management.”²⁴ We agree completely: the more accurate our comprehension of patient–ventilator interactions, the greater the benefit for our patients and for future technological developments. Although it is still hard to imagine that technology could replicate tasks or expert thinking that require human intelligence for the complex management of critically ill patients receiving mechanical ventilation, great progress is being made in that direction. Experience will always be valuable, but the future will be better for patients, families, and clinicians.