Control of Breathing During Mechanical Ventilation: Who Is the Boss?

Basics of Control of Breathing

The principles of control of breathing remain under-represented in the training curriculum of respiratory therapists, medical students, and even pulmonologists in training. While this may seem inexplicable, this is quite expected in that there is an explosion of knowledge and information in medicine, and there are many items competing for space and time in such curricula. Moreover, imparting theoretical concepts without adequate applied physiology may be an additional hurdle. This would be similar to teaching theoretical physics at a course entitled “Introduction to Engineering.” This part of this review is an attempt to provide the very basics of control of breathing with applied physiological exemplars. An additional goal of these exemplars is to underscore the stark differences of some of the basic assumptions of control of breathing in ambulatory subjects versus critically ill patients receiving mechanical ventilation.

Respiratory system control comprises the sensors, effectors, and the central controller (Fig. 1) that work together to ensure adequate gas exchange—intake of O₂ and elimination of CO₂, which, in turn, ensures normal pH for appropriate functioning of enzymes and other proteins that maintain cellular function. The central controller resides in the respiratory centers of the brain stem (medulla and pons). The respiratory centers comprise the inspiratory-firing and expiratory-firing neurons that reside in the ventral respiratory group of the ventrolateral brainstem. The ventral respiratory group is a collection of 3 nuclear groups: the Bötzinger complex, nucleus ambiguous, and nucleus retro-ambigualis. The neurons in the ventral respiratory group may be considered the upper motor neurons of the respiratory system and effect respiratory muscle activity by innervating and stimulating the motor neurons in the anterior horns of the spinal cord at various levels: cervical or thoracic for effecting activation of the upper airway, diaphragm, and intercostal muscles. The dorsal respiratory group, which lies in the dorsomedial aspect of the medulla oblongata receives afferent input from the peripheral chemoreceptors and pulmonary sensory inputs through the vagus and glossopharyngeal nerves.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

Control of breathing in humans. The central controller resides in the brain stem. The various components and stages of the effector arm (shown on the right) and the sensor arm (shown on the left) constitute a continuous feedback loop, with the end-product being gas exchange. Factors that influence each of these steps in this loops are shown within boxes. V̇_E = minute volume. V̇/Q̇ = ventilation/perfusion ratio.

In healthy subjects the respiratory controller ensures adequate blood gas exchange, which remains remarkably stable despite wide variations of ventilation. In contrast, the conventional paradigms of respiratory control may not apply to critically ill patients. In spontaneously breathing individuals the respiratory controller achieves normalization of gas exchange, remains responsive to physiological demands (eg, exercise and speech), and strives to ameliorate dyspnea. In patients receiving mechanical ventilation, however, gas exchange may take lower precedence to prevention of lung injury.

Exemplar 1: Respiratory Controller Versus Preventing Lung Injury

A 24-year-old woman with status asthmaticus is admitted with acute hypercapnic respiratory failure (P_aCO2 59 mm Hg, P_aO2 88 mm Hg, and pH 7.31 on room air). Following intubation, to prevent barotrauma, her arterial pH is allowed to fall to 7.14, with P_aCO2 at 98 mm Hg, while low tidal volumes of 6 mL/kg ideal body weight are administered in a volume-controlled continuous mandatory ventilation (VC-CMV) mode. Large doses of sedatives and analgesics are administered to counter the normal respiratory controller's response of tachypnea (and air hunger), to prevent breath stacking and worsening intrinsic PEEP. Despite heavy sedation, however, the young patient continues to exhibit an intact respiratory controller response of high inspiratory drive, and manifests double-triggering of the ventilator, with plateau pressure exceeding 35 cm H₂O. To prevent barotrauma, she is paralyzed with cisatracurium bolus and infusion, and heavy sedation is continued. pH is maintained at 7.25 with intravenous bicarbonate infusion, while the P_aCO2 is allowed to climb to 108 mm Hg with a P_aO2 of 102 mm Hg, on an F_IO2 of 0.40.

While the exemplar may be considered an extreme scenario, the reader should be cognizant that conventional paradigms (normalization of pH, arterial gases, and dyspnea) that govern control of breathing during unassisted breathing are significantly altered by the resetting of ventilation endpoints, administration of intravenous bicarbonate, sedation, analgesics, and even paralytic agents. However, the core principles of control of breathing will come back into play after a period of stabilization, when liberation from mechanical ventilation and initiation of spontaneous breathing is contemplated.

The sensors of the respiratory control system consist of the central and peripheral chemoreceptors and neuro-mechanical afferent inputs. The central chemoreceptors are located near the ventral-lateral surface of the medulla, in contact with the cerebrospinal fluid (CSF). The central chemoreceptors are stimulated by the H+ concentration in the CSF. The CSF concentration of H+ is regulated by P_aCO2, local blood flow in the area of the brain, and local metabolism. The blood/brain barrier prevents the H+ from the blood from gaining direct access to the medullary chemorecptors, whereas the blood/brain barrier is permeable to CO₂, which diffuses across the blood/brain barrier and combines with H₂O to liberate H+ ions, thereby stimulating the central chemoreceptors. Such an indirect means of influence of arterial pH on CSF pH (H+ concentration) has clinical importance.

Exemplar 2: Spurious Ventilator Dependence Due to Exuberant Ventilation

A 64-year-old man with COPD on home oxygen is admitted with acute hypercapnic and hypoxic respiratory failure (P_aCO2 79 mm Hg, P_aO2 58 mm Hg, and pH 7.21 on room air). His resting arterial blood gases on room air a year before were P_aCO2 54 mm Hg, P_aO2 58 mm Hg, and pH 7.36. He is obtunded and therefore intubated and placed on VC-CMV. Three days later, while still on VC-CMV, his arterial blood gases are P_aCO2 33 mm Hg, and P_aO2 112 mm Hg, and pH 7.52, on F_IO2 0.30, and he is mildly sedated. The therapist reports that attempts at spontaneous breathing trials (on pressure support of 5 cm H₂O and PEEP of 5 cm H₂O) have met with long apneas, followed by initiation of apnea ventilation. The therapist deems the patient not ready to wean.

In this exemplar, the chronic elevation of P_aCO2 would have prompted compensatory generation of CSF bicarbonate and resetting of CSF pH to normal, despite the elevated P_CO2 in the CSF and arterial blood. Subsequently, the aggressive ventilation and iatrogenic respiratory alkalosis would have substantially reduced H+ in the CSF (that already had an excess of bicarbonate anions), resulting in abnormally low ventilatory drive. Such reduced ventilatory drive, combined with sedation-induced lack of wakefulness drive to breathe, would result in central apneas. The occurrence of such central apneas can be misconstrued as ventilator dependence.

The peripheral chemoreceptors are located in the carotid bodies (at the bifurcation of the common carotid artery) and the aortic bodies are in the aortic arch. The carotid bodies are responsive to hypoxemia and become activated starting at a P_aO2 less than 100 mm Hg, whereas the aortic bodies respond to reductions in arterial pH, regardless of whether they are of respiratory or metabolic source.

Pulmonary stretch receptors, irritant receptors, and J receptors constitute the neuromechanical input to the respiratory controller. The pulmonary stretch receptors are within the airway smooth-muscle cells and are activated by stretch consequent to lung inflation. Such lung inflation initiates the Hering-Breuer inflation reflex, which prolongs expiratory time and thereby slows respiration. The irritant receptors and the J receptors play a role in lung response to noxious stimuli (cough) and pulmonary edema (rapid shallow breathing), respectively.

The effectors of the respiratory system begin with the upper motor upper neurons that are housed in the dorsal respiratory group in the medulla, and lead on to the lower motor neurons in the anterior horn cells of the spinal cord. These nerve impulses activate the respiratory muscles (upper airway, intercostal, and diaphragm muscle), which, in turn, generate negative intrathoracic pressure. The intrathoracic pressure overcomes the resistive and elastic forces of the respiratory system to achieve effective ventilation. This entire loop that is depicted in Figure 1, responding to a derangement in P_aCO2 (stimulus), leading to stimulation of the controller, and resulting in augmented ventilation through the effectors would constitute the integrated response of the respiratory system to CO₂.

The response of the respiratory system to changes in P_aCO2 or P_aO2 during wakefulness have certain characteristic features. During wakefulness, decrements in arterial oxygen saturation would lead to stimulation of the respiratory controller and consequent increments in minute ventilation (Fig. 2). In Figure 2 the increments in minute ventilation, even prior to an oxygen saturation of 90%, may be attributable to oxygen sensing by glomus cells of the peripheral chemoreceptors in the carotid bodies. Similarly, minute ventilation can increase by 2–3 L/min with a 1 mm Hg increase in P_aCO2 (Fig. 3). Such a ventilatory response is responsible for maintaining P_aCO2 and arterial pH in the desired range for the optimal functioning of the cellular proteins and enzymes. Such a homeostatic response is vital for any organism.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

Integrated response of the respiratory controller to changes in hypoxic stimuli, depicted as ventilatory response to changes in oxygen saturation. Change in ventilatory response during wakefulness is depicted by the black line. Changes in ventilatory response during slow-wave sleep, rapid-eye-movement (REM) sleep, and non-REM sleep are depicted by gray lines. Note the rightward shift and slope decrease of the ventilatory response when transitioning from wakefulness to sleep.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

Integrated response of the respiratory controller to changes in hypercapnic stimuli, depicted as ventilatory response to changes in P_CO2. Changes in ventilatory response during wakefulness are depicted by black lines. Changes in ventilatory response during various sleep stages are depicted by gray lines. Note the rightward shift and slope decrease of the ventilatory response when transitioning from wakefulness to sleep. Also note the interplay of hypoxic and hypercapnic stimuli, characterized by leftward shift and steeper slope of the P_CO2 response curve. The circle and square represent the operating point during calm resting breathing during wakefulness in normoxic (circle) and hypoxic (square) conditions. The vertical dashed arrows depict change in operating conditions when there is sudden sleep onset.

State-Dependent Changes

During sleep, however, there is a reduction of the magnitude and slope of the ventilatory response to hypoxia and hypercapnia (see Fig. 3). Such sleep/wakefulness state-dependent differences are evident during wakefulness as well. Specifically, despite a reduction of P_aCO2 below 36 mm Hg in Figure 3, the minute ventilation remains stable at about 5 L/min, exemplifying that there is an inherent wakefulness drive to breathe. Conversely, during sleep there is no such horizontal line or “hockey-stick” appearance of the CO₂ response curve (ie, wakefulness drive to breathe). Therefore, in a patient receiving mechanical ventilation, when such wakefulness drive to breathe is removed by the onset of sleep, if a patient were to be hypocapnic and there were no backup rate, then a central apnea may ensue.^2,3

Exemplar 3: Apnea During Invasive Mechanical Ventilation

A 54-year-old man recovering from cardiogenic pulmonary edema secondary to acute myocardial infarction is awaiting cardiac catheterization in the morning. Although he is intubated and mechanically ventilated, he is alert but agitated as he has difficulty breathing with the machine. The respiratory therapist changes him from VC-CMV to pressure support of 15 cm H₂O and maintains the PEEP and F_IO2 at 8 cm H₂O and F_IO2 0.50. The patient becomes less agitated and appears more comfortable and is generating tidal volumes between 700 mL and 900 mL. At 10:00 pm he receives an aliquot of midazolam for nighttime sedation, to ensure adequate rest. The nighttime nurse notices recurrent ventilator alarming and notes that the apnea ventilation has been triggered off and on. The patient is placed back on VC-CMV and the apnea alarms are prevented.

In this exemplar, the patient with heart failure and J-receptor stimulation is more likely to be hypocapnic, and with sleep onset in the absence of a backup rate (pressure support only), the central apneas ensued. Previous studies demonstrated that apneas during mechanical ventilation are not unique to healthy subjects receiving assisted ventilation under experimental conditions, but can occur in critically ill patients receiving mechanical ventilation.^3–5 However, the clinical implication of the occurrence of such central apneas during mechanical ventilation is unclear. Conceivably, such central apneas could disrupt sleep and lead to unfavorable stress responses or even neurological effects, such as delirium.^6–8

While the occurrence of central apneas are gross changes in patient-ventilator interactions in response to sleep/wakefulness state changes, more subtle changes can occur as well. For instance, inspiratory time can increase by 23% and expiratory time can increase by 126% with the change from wakefulness to sleep,³ which would reduce the respiratory rate, which during pressure support ventilation is used as a measure of respiratory unloading. Conceivably, clinical decisions, such as determination of the pressure-support level, could be influenced by the sleep/wakefulness state when the physician is setting the ventilator. This is particularly important when more profound changes are anticipated, as in the exemplar below.

Exemplar 4: Going to Surgery

A frail 33-year-old woman is admitted with severe acute respiratory distress syndrome and is receiving low-tidal-volume mechanical ventilation. Her ventilator settings are VC-CMV with a backup rate of 14 breaths/min, a spontaneous rate of 32 breaths/min, tidal volume of 360 mL, F_IO2 of 0.70, and PEEP of 12 cm H₂O. Her arterial blood gas values are pH 7.25, P_aCO2 44 mm Hg, and P_aO2 71 mm Hg. Serum chemistries reveal a large anion-gap metabolic acidosis, elevated serum lactate, and acute renal insufficiency (serum creatinine 2.1 mg/dL and serum potassium 4.5 mEq/dL). The patient is easily arousable and cooperative, and grimaces when her abdomen is touched. She is quickly diagnosed with a perforated duodenal ulcer and she needs to undergo emergency laparotomy. She is taken to the operating room and general anesthesia is induced easily, as she was already intubated. Before surgery can commence, she develops profound bradycardia and severe hypotension with broad QRS complexes, and arterial blood gas analysis reveals pH 6.95, P_aCO2 69 mm Hg, and P_aO2 61 mm Hg, and serum potassium concentration is 6.2 mEq/dL.

State-dependent changes in control of breathing that occur with deep sedation during anesthesia could lead to life-threatening situations. In this exemplar the patient's spontaneous respiratory rate was 32 breaths/min, despite a machine-set (backup) rate of 14 breaths/min. Following induction of anesthesia, her respiratory rate fell from 32 breaths/min to 14 breaths/min, with worsening of respiratory acidosis (compounded by large dead space in this patient with acute respiratory distress syndrome) due to the loss of wakefulness drive to breathe and CO₂ responsiveness following induction of anesthesia.

Other mechanisms besides chemical influences may play a role in the state-dependent changes in ventilation. In critically ill patients receiving pressure support, an increase in tidal volume, achieved by small increments in pressure assist, prolonged (neural) expiratory time and decreased respiratory rate.⁹ Such a prolongation of expiratory time in response to lung inflation (the Hering-Breuer reflex) is known to be operational during sleep and not during wakefulness, and is mediated by the neuromechanical sensors of the respiratory control system (see Fig. 1).¹⁰

Interplay of Stimuli

The integrated system responses to alteration of stimuli (such as hypercapnia or hypoxia) are more than additive.¹¹ In Figure 4, during hypoxia the ventilatory response to increments in P_aCO2 is shifted to the left and is steeper than that during normoxia. Such a shift in the ventilatory response can result in ventilatory instability, which was exemplified in exemplars 3 and 4 above. Such complexity in the respiratory system's operation can be used as a schematic to model and test responses to interventions and therapy. The stability of the integrated response of the respiratory system to ventilatory stimuli, depicted in Figure 1, can be summed up as loop gain. Loop gain is essentially an engineering term that describes the stability of a system controlled by feedback loops, such as the human respiratory system. A loop gain close to zero indicates stability, whereas a loop gain close to, or greater than, one indicates an unstable system.¹² Various factors besides the coexistence of hypoxia can destabilize the respiratory controller. Specifically, arousals from sleep, rapid transitions from wakefulness to sleep state (as in exemplar 4), hypocapnia, and administration of pressure assist such as pressure support could augment the CO₂ responsiveness slope and thereby increase loop gain and ventilatory instability.¹¹ Moreover, slow circulation (eg, in a patient with congestive heart failure, as in exemplar 3) could create delays in the respiratory controller's response “sensing,” and therefore response to changes in P_aCO2 could cause oscillations in the system and ventilatory instability.⁵

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

Integrated response of the respiratory controller to changes in hypercapnic stimuli, depicted as ventilatory response to changes in P_CO2. Changes in ventilatory response during wakefulness are depicted by black lines. Changes in ventilatory response during various stages of sleep are depicted by gray lines. Note the interplay of hypoxic and hypercapnic stimuli, characterized by leftward shift and steeper slope of the P_CO2 response curve, and how various other factors (eg, PEEP, F_IO2, sedatives, acetazolamide, and external CO₂ administration) can stabilize the respiratory controller, whereas hypoxia, pressure support (PS), and sedatives can destabilize the respiratory controller. The circle and square represent the operating point during calm resting breathing during wakefulness in normoxic (circle, point B) and hypoxic (square, point A) conditions. The vertical dashed arrows depict change in operating conditions when there is sudden sleep onset. Note how the sudden onset of sleep at point A could result in minute ventilation of zero (central apnea), whereas a similar situation at point B would result in reduced ventilation but not apnea.

Alternatively, ventilatory instability can be improved by administering acetazolamide, or increasing CO₂ by increasing dead space, or administering CO₂ gas.^3,13 Moreover, correcting hypoxia by increasing F_IO2 or PEEP could attenuate the steep slopes in the ventilatory response curves to hypercapnia, and thereby stabilize breathing. For example, in the presence of hypoxia a patient could be operating at point A in Figure 4. If this patient were to suddenly fall asleep, then ventilatory response would fall to 0 L/min, resulting in a central apnea. In contrast, if the patient were not hypoxic, the patient would have been operating at point B in Figure 4, and sudden sleep onset would decrease ventilation but not cause an apnea. Also, reducing the pressure assist (pressure support) level could prevent such ventilatory instability and apneas.^4,14,15 Such an integral knowledge of applied physiology could help the bedside clinician troubleshoot the problem and take corrective action.

Three Pumps

In general, during mechanical ventilation the relationship between the respiratory system and the mechanical ventilator is considered to be 2 pumps working in concert. However, a commonly overlooked layer of complexity is that during mechanical ventilation there are actually 3 pumps that need to be working in synchrony: the ventilator, and the inspiratory and expiratory muscle pumps (Fig. 5). The complex respiratory controller, replete with chemical and neural “servo” inputs and cortical inputs, is attached to the mechanical ventilator, which is controlled by a rudimentary set of microcircuitry. The interaction between these 2 systems, 3 pumps, and 2 controllers—one servo-operated and the other rules-based with lack of any relevant feedback (servo) loop—is a clear mismatch and needs rethinking. Why do we need to study the interaction between the respiratory controller and the circuitry and algorithms that control the ventilator pump? The study of the interaction between these 2 systems and 3 pumps is vital to ensure adequate patient-ventilator synchrony, which, in turn, would ensure adequate respiratory muscle rest, gas exchange, and patient comfort. For these goals to be achieved, we need to have an in-depth understanding of respiratory control during mechanical ventilation, and more importantly, ultimately we need to model the circuitry and algorithms of the mechanical ventilator to parallel that of the respiratory controller.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 5.

The interaction between 3 pumps (inspiratory, expiratory, and ventilator) acting on the same column of air, that are differentially controlled by a complex feedback-based respiratory neural controller in humans, as opposed to a simple bank of microcircuitry of the mechanical ventilator that is devoid of (chemical or neural) feedback loops.

The interaction between the patient and the ventilator can be dissected during the course of a single breath and over multiple breaths in order to better understand control of breathing as it pertains to patient-ventilator interaction. Over a single breath, as shown in Figure 6, the breath could be broken into 5 time segments: the trigger phase (ie, the switch from expiration to inspiration); the post-trigger phase; the inspiratory phase; the cycle phase (ie, the switch from inspiration to expiration); and the expiration phase.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 6.

The phases of a breath during mechanical ventilation.

Delay of triggering can result in a delay between when the patient's (respiratory controller) effort commences and when the ventilator's inspiratory valve is triggered,¹⁶ which can increase respiratory effort, and which could spill into the post-trigger and inspiration phases (Fig. 7).^17–19 While flow triggering could reduce the inspiratory effort during triggering (compared to pressure triggering), in the post-trigger phase the inspiratory effort is comparable for flow and pressure triggering. Research on newer methods for predicting and timing inspiratory triggering, by going further upstream in the effector arm of the respiratory controller (see Fig. 1), is ongoing, and some of these devices are available for patient care. For example a noninvasive “shape signal” method, wherein the shape of the flow curve during the tail end of expiration is used to predict the next inspiratory effort, has been used in mechanical ventilation of critically ill patients.²⁰ Such efforts have been preceded by similar tracking of flow mechanisms for bi-level positive airway pressure ventilation in ambulatory patients with sleep-disordered breathing, to ensure adequate cycling of bi-level devices.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 7.

The juxtaposition of the patient's respiratory cycle (inner ring) versus the corresponding phases of the mechanical ventilator breath (outer ring) can be depicted by this dial-within-a-dial concept. Left: Perfect patient-ventilator synchrony. Right: Phase lag in mechanical cycles that can be easily conceptualized as a slight turn of the outer (ventilator) dial. Note how the mismatch in the inspiratory phase between the patient and the ventilator spills over to the exhalation phase (red arrowhead): the patient has begun exhalation but the mechanical ventilator is still in its inspiration phase. Such a failure to terminate inspiration (ie, cycle delay) could lead to dynamic hyperinflation.

Changes in transdiaphragmatic pressure or diaphragm excitation would be further upstream in the effector arm of the ventilator circuitry (see Fig. 1), but these techniques are invasive and require the placement of an esophageal probe.^21,22 Nevertheless, such techniques can reduce triggering and cycling delays and ensure adequate patient comfort, even in awake patients who are receiving invasive mechanical ventilation.²¹ Moreover, such techniques would ensure adequate ventilatory support during inspiration.²³ Such instantaneous feedback from the respiratory controller to the mechanical ventilator would ensure an optimal balance between patient effort and ventilatory assist, after overcoming the work load posed by respiratory resistance and elastance.²⁴

Delays between the onset of the patient's inspiratory effort and respiratory controller and mechanical inspiration could lead to asynchrony at the expiratory-to-inspiratory switch point (see Fig. 7), because the expiration time is truncated by the delay in opening of the expiratory valve.¹⁶ Experimental manipulation of the ventilator settings to induce such a delay have been shown to induce asynchrony at the expiratory-to-inspiratory switch point, which, in turn, could lead to dynamic hyperinflation.²⁵ Such dynamic hyperinflation, in turn, can lead to non-triggering effort of the subsequent breath in experimental human models as well as real-life circumstance.^16,26 The non-triggering efforts in turn could lead to patient discomfort. However, while the rationale for achieving adequate synchrony between the patient's respiratory controller and the mechanical ventilator is strong, there are currently no outcomes-based studies to confirm the benefit of this approach.

Who Is the Boss?

A key question when designing closed-loop systems is to conceive the parameters and boundaries that will guide the settings and the functioning of the ventilator. Specifically, the question remains as to who is right: the patient or the physician's notion of what is ideal for the patient? (Fig. 8)

• Download figure
• Open in new tab
• Download powerpoint

Fig. 8.

The pyramid depicts concepts that are integral to our understanding of control of breathing in the context of mechanical ventilation. The basic understanding of the motor and sensory components of the respiratory controller form the foundation of this understanding, but sleep-wakefulness state-dependent changes, interplay between various respiratory stimuli, the nature of three (inspiratory, expiratory, and ventilator) pumps acting on the same column of air, and the effects of time compound and nuance our understanding of respiratory control during mechanical ventilation. The identical triangle in the background reflects a parallel universe, and much ignored, or concept of what the patient desires versus the triangle in the forefront that reflects the providers' understanding and construct of control of breathing during mechanical ventilation. Such a paradigm can overturn some of the assumptions a provider or researcher makes on a day-to-day basis.

Specifically, in determining the optimal ventilatory strategy, is allowing the patient's intact respiratory controller to respond to a given clinical state paramount, or is the avoidance of the adverse effects of ventilatory effort on respiratory muscle more important? And are there trade-offs? And, if so, how much do we unload or load a patient? These gnarly questions need to be untangled so as to allow us to advise the closed-loop system appropriately. The above dilemma could be better understood with the following exemplar.

Exemplar 5: To Unload or Not to Unload: That Is the Question

A 42-year-old man is recovering from status asthmaticus and steroid-induced myopathy. His oxygenation and ventilatory mechanics are adequate, but, due to profound weakness, he has repeatedly failed attempts to liberate him from mechanical ventilation. Between the spontaneous breathing trials, the patient is placed on VC-CMV. But a new episode of ventilator-associated pneumonia and acute respiratory distress syndrome requires deep sedation and low-tidal-volume ventilation.

Passive mechanical ventilation is associated with marked atrophy of the diaphragm and increased diaphragmatic proteolysis, so in clinical practice we try to avoid suppressing the patient's spontaneous respiratory effort.³² In this “difficult-to-wean” patient with profound respiratory muscle weakness, the administration of VC-CMV and deep sedation may lead to periods of passive mechanical ventilation that are difficult to monitor and adjust-for at the bedside.^33,34 Animal studies have not clearly defined the minimal duration of controlled mechanical ventilation that invariably results in ventilator-induced diaphragmatic dysfunction.³⁵ Therefore, state-dependent changes in respiratory rate may lead to excessive respiratory muscle resting and resultant muscle atrophy.³² In exemplar 5, both VC-CMV and pressure support were valid options, and there are animal data to suggest that pressure support is less likely to cause muscle atrophy.³⁴ But pressure support may be associated with sleep-related central apneas and sleep disturbances,³ so the ideal ventilation settings would be pressure support with a backup rate, which would sense patient-triggered versus machine-mandated breaths and guarantee a certain proportion of patient-triggered breaths.

Similarly, the choice of tidal volume may be influenced by the state of the respiratory controller: sleep or wakefulness.³⁰ In patients with chronic respiratory insufficiency receiving averaged volume-assured pressure support, tidal volume was chosen with 2 methods: one based on patient preference (110% of ambient tidal volume during calm wakefulness), and another based on ideal body weight (8 mL/kg ideal body weight) (Fig. 9). The patients preferred (as measured via dyspnea) 110% of ambient tidal volume during wakefulness, but the patients slept better (as measured via sleep efficiency) when receiving tidal volume based on ideal body weight.³⁰ On the one hand, this state-dependent change in the respiratory controller can significantly impact intermediate physiological endpoints such as sleep and dyspnea. Moreover, such data would suggest that ventilators that can sense sleep/wakefulness may be warranted. More importantly, such data further underscore that it is unclear as to who knows what ventilator setting is best: the patient or the provider.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 9.

State-dependent and patient-dependent differences in intermediary physiological end points. In patients with chronic respiratory insufficiency receiving averaged volume-assured pressure support, tidal volume was administered based on 2 different methods: one based on ideal body weight (8 mL/kg IBW), the other based on patient preference (110% of ambient tidal volume during calm wakefulness). The patients preferred (as measured by subjective dyspnea ratings) 110% of ambient tidal volume during wakefulness (left), but the patients slept better (as measured by polysomnography-based sleep efficiency) (right) when receiving tidal volume based on ideal body weight. (Adapted from Reference 30.)

Future Directions

Market forces are a factor that needs to be dealt with when contemplating new computerized “intelligent” systems that enable adequate mechanical ventilation while addressing the control-of-breathing issues. Whereas the sleep medicine field has benefited from a larger sales market, the same has not been the case for mechanical ventilation in the acute critical care unit. Another obstacle is the regulatory concern over closed-loop systems and the lack of clear scientific data that would guide ventilator settings in different patient populations. While the current focus of patient-ventilator interactions in the context of control of breathing addresses intermediate physiological endpoints such as dyspnea and gas exchange, more long-term studies that measure tangible outcomes such as ventilator dependence and mortality are needed.