“Elastance” (E) describes a behavior of a system in which a certain change in pressure occurs necessarily with a given change in volume. It can be measured in units of cm H₂O per mL. Lung elastance (E _l) causes a rise in lung elastic recoil pressure per unit of tidal volume gained. During a normal inspiration, a decreasing pleural pressure needs to be generated by the respiratory muscles to overcome the rising lung elastic recoil pressure. The inspiratory rise in elastic recoil pressure can be opposed by a rise in positive airway pressure. If the rise in positive airway pressure is applied by a ventilator such that it increases in proportion to the tidal volume, the ventilator itself exerts an “elastic behavior” characterized by the change in ventilator pressure per unit of tidal volume. This elastance of the ventilator (E _v) is equally quantifiable in cm H₂O/mL. However, while lung elastic recoil pressure tends to collapse the lung, the ventilator pressure acts in the opposite direction in the elastic unloading mode. Relative to lung elastance, the elastic “behavior” of the ventilator is therefore “negative.” When a patient is connected to a ventilator in the elastic unloading mode, the overall elastance of the combined system of lung and ventilator is equal to the arithmetic sum of the two elastic elements (because they are interconnected “in series”):
with E _v being a negative term.

The respiratory muscles work against this overall elastance of the combined system which becomes lower than lung elastance itself when the negative elastic behavior of the ventilator is added. If ventilator elastance is adjusted such that the overall elastance approaches the magnitude of a normal lung’s elastance, respiratory muscles will work on a system of apparently normal elastance. It follows that the patient’s elastic work of breathing will then be in the normal range while the disease-related part of the elastic work of breathing is eliminated by the ventilator. To calculate the magnitude of ventilator elastance needed to achieve a “normal” overall elastance, the equation has to be solved for E _v:

For example, an infant weighs 750 g and has a measured respiratory system compliance of 0.3 mL/cm H₂O. The following steps are required to calculate precisely the gain of elastic unloading that returns the infant’s elastic work of breathing to normal:

Thus, if the ventilator is set to raise the airway pressure by 2.66 cm H₂O for each mL of tidal volume, this infant’s elastic work of breathing would be in the “normal range.”

This example shows that the required level of elastic unloading (gain) in cm H₂O/mL depends not only on the absolute stiffness of the lungs but also on body weight. Smaller infants will require higher gains. This is because “normal” lung compliance varies directly with body weight. Because the normal tidal volume in infants also varies directly with body weight, the respiratory muscle pressure amplitude required to generate a normal tidal volume is relatively constant across different body weights.

For clinical purposes, it is sufficient to base the estimate of the required elastic unloading gain on measurements of total respiratory system compliance rather than lung compliance. The former variable is easier to measure because it does not require esophageal pressure manometry. Total respiratory system compliance is not very different from lung compliance in premature infants because of their soft chest wall (high chest wall compliance).

Similar to elastic unloading, the applied ventilator pressure waveform during resistive unloading is in close relation to the airflow of spontaneous breathing, and this “behavior” can be described as a “negative resistance,” i.e., a certain change in applied pressure per unit of airflow (cm H₂O of ventilator pressure per L/s). For a patient on a ventilator, the overall resistive load on respiratory muscles is the sum of the different components:
with R _aw being the airway resistance, R _ETT the endotracheal tube’s resistance, and the ventilator’s resistance R _v, the latter being a negative term.

For example, an infant is intubated with a 2.5-mm inner diameter endotracheal tube. This endotracheal tube has an estimated resistance of 20 cm H₂O/L/s. If the gain of resistive unloading is set at – 20 cm H₂O/L/s, the infant can breathe as if the endotracheal tube’s resistance would not be in place. The ventilator specifically relieves the amount of additional resistive work of breathing associated with this endotracheal tube.

Impairment of respiratory mechanics in infants is associated with increased chest wall distortion (CWD). CWD further increases respiratory muscles’ work of breathing because work is “wasted” in distorting the chest. This reduces the efficiency of the respiratory muscles’ to generate ventilation. Compensating for impaired respiratory mechanics by unloading decreases CWD and therefore improves the efficiency of respiratory muscles. This has been demonstrated in small adult and newborn animals who were intubated and had their lungs injured with meconium (Schulze et al. 1998b) and in very low-birth-weight infants with acute respiratory illness (Musante et al. 2001). These studies have also shown that compared to unassisted breathing, a larger fraction of the tidal volume enters the chest compartment while the displacement of the abdominal compartment does not change much with respiratory mechanical unloading. Unloading can therefore provide a “stenting pressure” and stabilize the preterm infant’s highly flexible chest wall during spontaneous breathing.

Respiratory center output is increased in lung disease with impaired respiratory mechanics. It has been shown in animal models with severe pulmonary parenchymal failure that PAV/RMU decreases phrenic nerve activity while tidal volume rises and blood gases improve (Schulze et al. 1999b).

Magnitudes of unloading that equal or surpass lung elastic recoil (or pulmonary resistance) may induce a runaway of the ventilator pressure. In such situation, a small initial change in volume (airflow) triggers an increase in ventilator pressure higher than the accompanying rise in lung elastic recoil (flow-resistant airway pressure gradient) which in turn further increases lung volume (airflow). This positive feedback self-perpetuates because the negative elastance of the ventilator exceeds the positive elastance of the respiratory system. With such excessive gain settings, artifacts on the volume/airflow signal may initiate ventilator pressure runaway up to the set ventilator pressure limit or a set tidal volume limit. Unless ventilator pressure limits are in place, overcompensation of lung elastic recoil (excessive gains of volume-proportional assist) will lead to a self-perpetuating passive inflation. This process will not be terminated until overinflation leads to stiffening of the lung (increased lung elastance, decreased lung compliance) so that the ventilator pressure finally equals lung elastic recoil pressure (lung elastance equals the magnitude of the set “negative” ventilator elastance). As lung recoil pressure is completely balanced by the ventilator pressure at this point, passive expiration will still not ensue even though the inspiratory effort may be long over. Therefore, besides tidal volume limits, an airway pressure limit and an inflation time limit should always be in place during PAV/RMU (Schulze et al. 1998a).

Full compensation of pulmonary resistance (when negative ventilator resistance equals pulmonary resistance in magnitude) induces sinusoidal oscillations of the respiratory system near its resonant frequency. This phenomenon can be explained by electrical or mechanical analogies of the respiratory system. The amplitude of the oscillations increases with higher gains of resistive unloading so that a variant of high-frequency oscillatory ventilation near the resonant frequency may be generated (Schulze et al. 1991).

A leak around the endotracheal tube will be measured as inspiratory airflow and volume by the pneumotachograph that is mounted at the endotracheal tube adapter. In the PAV/RMU modes, this will cause a rise in airway pressure out of proportion to the airflow and volume entering the lung. In case of a substantial leak, the ventilator pressure will rise rapidly to the set pressure limit and then fall to the PEEP level and may rise again repetitively. However, the leak component of the airflow signal can be estimated by the ventilator so that the airway pressure changes can then be based on a corrected flow signal (Fig. 8.26). With appropriate algorithms, PAV/RMU devices for infants can perform satisfactorily in the presence of variable leaks of up to 20–30 % of the tidal volume.

The basic principle of ASV stems from the work by Otis et al. (1950) and Mead (1960) demonstrating that, for a given level of alveolar ventilation, there is a particular RR that minimize the WOB. This optimal RR is the best compromise between elastic WOB and resistive WOB (Fig. 8.27) and depends on the minute volume (MV), the expiratory time constant (RC_exp), and the dead space (V _D). To initiate ASV, the clinician sets the patient’s predicted body weight (PBW) and the percentage of minute ventilation (100 % being equal to 100 mL/kg PBW/min in adult patients). V _D is computed by the Radford monogram at 2.2 mL/kg PBW. The ventilator then delivers five test inflation to determine RCexp by analysis of the expiratory flow–volume curve (Brunner 2001; Brunner and Iotti 2002; Laubscher et al. 1994b; Brunner et al. 1995; Lourens et al. 2000). Thereafter, the ventilator compute the target V _T and RR, and mechanical ventilation starts using an interbreath negative control of V _T and RR to drive the patient’s breath pattern to the target V _T and RR (Brunner 2002). In passive patients unable to trigger a breath, the ventilator generates pressure-controlled inflation and automatically adjusts the inspiratory pressure (P _insp) and timing to achieve the target V _T and RR. In active patients able to trigger a breath, the ventilator switch to pressure support inflation, automatically adjusting P _insp to achieve the target V _T, and it delivers additional pressure-controlled inflation if the patient’s RR is below the target RR. During the maintenance, RCexp is estimated on a breath-by-breath basis to reassess the target and adjust P _insp, I:E ratio, and mandatory RR to maintain the target MV and RR within a frame designed to avoid both rapid shallow breathing and excessive inflation volumes. These safety limits to prevent markedly abnormal ventilator settings are automatically determined or can be manually adjusted. The limits are (min–max) as follows: inspiratory pressure (5 cm H₂O above PEEP to 10 cm H₂O below set P_max), V _T (4.4 mL/kg PBW to 22 mL/kg PBW), mandatory respiratory rate (5/min to 20/RC_exp/min), inspiratory time (0.5–2 s), expiratory time (3 RC_exp to 12 s), and inspiratory/expiratory time ratio (1:4–1:1). The rather high value of 22 mL/kg PBW for the upper limit of tidal volume can be set at a lower level to avoid lung overdistension by setting P _max.

At initiation, clinician has to set MV by setting PBW and %MV. It is crucial to set properly the PBW and not the actual body weight. A large discrepancy between PBW and set body weight may result in large V _T delivered (Dongelmans et al. 2008). Usually, %MV is set at 100 % as a starting point (110 % if an HME or heat and moisture exchanger is used on the ventilatory circuit). Later, %MV is adjusted by 10–20 % according to the PaCO₂ measured on blood gas analysis. In addition, settings of PEEP, FiO₂, P _max, inspiratory trigger sensitivity, rise time, and expiratory trigger sensitivity are required. In passive patient, monitoring is similar to the pressure control mode. Plateau pressure is estimated by P _insp if inspiratory flow reach zero at the end of inspiration, otherwise an end-inspiratory occlusion is required. Total PEEP is assessed by an end-expiratory occlusion maneuver, especially in asthmatic and COPD patients (Brochard 2002). In active patient, monitoring is similar to pressure support. P _insp required to reach the target V _T informs on the readiness to wean the patient. If the ASV screen shows that the patient’s RR is far above the target RR, it means a high ventilatory drive that may require increasing the %MV in addition to the etiologic treatment (Wu et al. 2004) (Fig. 8.28). In all cases, monitoring of RCexp inform of the respiratory mechanics. The normal value for intubated adult patient is around 0, 75 s (Arnal et al. 2008). A longer RCexp suggests an obstructive disease (asthma, COPD, bronchospasm, etc.) if not an excess of secretion or an endotracheal tube kicking. A shorter RCexp suggests a restrictive disease either due to the lung (atelectasis, ARDS, etc.) or to the chest wall (morbid obesity, kyphoscoliosis, etc.).

A multicenter trial tested the safety of ASV as compared to volume control mode (VC) in 48 passive adult patients (Iotti et al. 2005). In term of peak pressure, Pplat, and intrinsic PEEP, ASV was at least as safe as VC. Furthermore, in one-third of these patients, ASV was more effective than VC in clearing CO₂. An observational study in a cohort of 243 adult patients with different lung conditions reported different V _T–RR combination delivered according to the respiratory mechanics. V _T delivered was 8.3, 9.3, and 7.6 mL/kg PBW for normal lung condition, COPD, and ARDS condition, respectively (Arnal et al. (2008). In ARDS patients, a bench study compared V _T and Pplat delivered by ASV versus a strict 6 mL/kg PBW strategy (Suleymanci et al. 2008). Result showed that in the most severe cases of low compliance, high PEEP, and high MV, ASV was able to keep Pplat lower than the 6 mL/kg strategy. This is in favor of a good adaptation of the breath pattern delivered according to the respiratory mechanics with a V _T delivered below 6 mL/kg PBW for the most severe cases. A subsequent observational study focusing on 51 unselected ARDS patients confirmed the previous result with V _T delivered at 7.8 mL/kg PBW on the first day of mechanical ventilation correlated to the static compliance. In addition, Pplat was kept ≤28 cm H₂O during all the ventilation duration (Arnal et al. 2006). Two preliminary studies in large numbers of patient report the use of ASV in 98 % of ICU patients (Arnal et al. 2004) with a rate of failure of 3 % (Sviri et al. 2007). To resume, ASV can be used with safety in all conditions of passive ventilated ICU patient. The breath pattern selected is adapted to the respiratory mechanics and is in line with recommendations regarding protective ventilation strategy in ARDS patients (Yilmaz and Gajic 2008).

In stable patients spontaneously breathing, the ASV algorithm will progressively and automatically reduce the inspiratory pressure as the patient’s respiratory mechanics improve. A bench study determined that ASV effectively maintains minute ventilation and alveolar ventilation under various conditions of spontaneous ventilation, compliance, and airway resistances (Lawson et al. 2001). A physiological study in ten patients during early weaning with partial ventilatory support suggested an improvement in patient–ventilator interaction and reduced signs of dyssynchrony with ASV compared with SIMV–PS (Tassaux et al. 2002). Another physiological study in 14 adult patient during weaning phase showed that ASV and PSV behaved differently but ended up with similar pressure level facing acute changes in ventilatory demand by contrast to dual modes (volume-guaranteed pressure-control mode), in which an increase in ventilatory demand results in a decrease in the pressure support provided by the ventilator (Jaber et al. 2009).

Preliminary studies in adults suggest that ASV can simplify ventilator management and reduce the time to extubation in predominantly postoperative patient populations. Two RCTs involving 49 and 34 post-cardiac surgery patients, respectively, found that compared to protocols using SIMV and PS, ASV resulted in fewer changes in the ventilator settings from the ICU team providing care (Sulzer et al. 2001; Petter et al. 2003). Additionally, the first one noted that ASV resulted in significantly fewer high-inspiratory pressure alarms (Sulzer et al. 2001), while the second one also noted that ASV significantly reduced the duration of endotracheal intubation [3.2 (2.5–4.6) vs. 4.1 (3.1–8.6) h; P = 0.02] and resulted in more fast-track successes (successful extubation at 6 h) (Petter et al. 2003). Two recent RCT performed in post-cardiac surgery patients found conflicting results (Gruber et al. 2008; Dongelmans et al. 2009). The first one included 50 patients after elective coronary artery bypass grafting and found earlier extubation associated with the use of ASV, without an increase in clinical intervention, when compared with pressure-regulated volume-controlled ventilation (Gruber et al. 2008). The second one compared ASV with pressure support for the post-cardiac surgery weaning in 128 patients and did not find any benefit in terms of ventilation duration (Dongelmans et al. 2009). In all these study, the weaning protocol associated with the use of ASV interfere with the results.

To test the response of ASV in small pediatric patients with a sizeable endotracheal tube leak, a bench study found that MV delivered was not affected by leaks as high as 50 % for patients whose PBW is 5 kg (Watson et al. 1999). From 10 to 20 kg, MV delivered is affected if leaks are above 25 %. A short-term crossover designed study compared ASV with SIMV in 13 children (mean age 2.3 years, actual body weight 12.3 kg) (Parret et al. 2000). Both V _T and peak pressure were decreased with ASV associated with an increase in inspiratory time and a decrease in expiratory time. At the present time, ASV can be used in children in the same way as in adult. In neonates, the experience is too scarce to recommend its use.

Limitations. The absolute contraindication of ASV is an important leak on the circuit as the monitoring of the V _T would be disturbed. Thus, ASV is currently not recommended in noninvasive ventilation and in case of bronchopleural fistula. ASV is more and more used around the world both in adult and pediatric population. Still, data are lacking concerning the weaning duration in ICU (beside postoperative patients), workload, cost, and outcomes especially in pediatrics.

Unlike the conventional ventilatory modes which adapt the assistance to airway pressure or flow, NAVA is adjusted by the neural drive to the diaphragm. The key parameter is the EAdi, which is assumed to reflect the patient respiratory center demand. EAdi signal is continuously recorded using an array of nine miniaturized electrodes mounted on a conventional feeding tube positioned in the lower esophagus (Fig. 8.29). An automated method for online acquisition and processing of EAdi signal has been developed to obtain a high signal-to-noise ratio, to limit electrocardiogram influence, and to maintain an optimal muscle-to-electrode distance despite the diaphragm displacement. Briefly, a cross-correlation analysis of signals measured by each electrode pair permits to identify the diaphragm position (along the array): The most negative correlation value is obtained from the electrode pairs located on either side of the diaphragm (Beck et al. 1996; Sinderby et al. 1997). The signals given by the electrode pairs the closest to the diaphragm are subtracted and filtered, and the remaining heart artifacts are suppressed by replacement (Sinderby et al. 1995, 1998). The final EAdi signal is transferred to the ventilator.

EAdi reflects the respiratory center demand, as illustrated by the EAdi increase observed when the respiratory load is artificially augmented, or in patients with restrictive or obstructive respiratory diseases (Sinderby et al. 1998). Good quality EAdi signals have been successfully recorded in healthy volunteers, critically ill adult patients, pediatric patients including premature infants (Beck et al. 2009; Emeriaud et al. 2006), and small animal species (Allo et al. 2006). Adequate EAdi has also been recorded in patients with severe neuromuscular disease (Beck et al. 2006).

During NAVA, the ventilatory assist is triggered when EAdi exceeds a threshold level (Fig. 8.30). The magnitude of the mechanical support during inspiration is adapted every 60 ms, corresponding to the EAdi multiplied by a proportionality factor that can be selected, called the NAVA level. Inspiratory support is interrupted when EAdi decreases below 70 % of the peak EAdi, and the selected PEEP is then applied. Experiences of short periods (maximum 8 h) of ventilation with NAVA have been reported in healthy volunteers (Moerer et al. 2008), critically ill adult patients (Brander et al. 2009a; Colombo et al. 2008; Spahija et al. 2010), pediatric patients (Breatnach et al. 2010; Bengtsson and Edberg 2010), premature infants (Beck et al. 2009), and small animal species (Beck et al. 2008; Lecomte et al. 2009).

One major advance with NAVA concerns the patient–ventilator synchronization. The electrical activation of the diaphragm precedes its contraction and the modification of pressure or flow in the respiratory circuit; cycling-on delay is intrinsically lower with NAVA than with conventional ventilation based on pneumatic triggering. An important point is that this synchrony should not be affected by the presence of leaks nor by unfavorable mechanical conditions like overdistension with intrinsic PEEP.

The improvement in patient–ventilator synchrony has been confirmed in lung-injured rabbits, with lower trigger delay and fewer non-assisted inflation with NAVA than with pressure-support ventilation (PSV) (Beck et al. 2007). In adult critically ill patients, NAVA was associated with lower asynchrony than PSV, particularly at high level of assistance (Colombo et al. 2008; Spahija et al. 2010); shorter delays were also observed with NAVA for both triggering the assist and cycling off (Spahija et al. 2010). In 7 premature infants (weight 976 ± 249 g), trigger delays were also similar with NAVA or PSV, but the cycling off was more synchronous in NAVA (Beck et al. 2009). Importantly in the same study, the trigger and cycling-off delays in NAVA were not altered after extubation (Beck et al. 2009). In pediatric patients, the synchrony has been less evaluated, but it has been shown that the neural triggering precedes the pneumatic triggering in more than 2/3 inflation during NAVA (Breatnach et al. 2010; Bengtsson and Edberg 2010).

During noninvasive helmet ventilation, an improved patient–ventilator synchrony has also been demonstrated in volunteers (Moerer et al. 2008).

The patient–ventilator synchrony is also improved with NAVA in a second dimension: The assist is delivered proportionally in response to the patient respiratory center demand. This relation relies on the concept of unloading the diaphragm during the period of increased respiratory load, in order to prevent a diaphragm fatigue. This concept is quite different than other conventional modes used in pediatric and neonatal patients, like pressure support with volume guarantee, assist-control ventilation, and volume assist. With these modes, when the patient efforts augment, the ventilator progressively decreases its assistance after a few cycles. This can be considered an indication of weaning readiness, but this can also augment the diaphragm fatigue or the oxygen consumption if the situation is prolonged. In contrast, an increase in EAdi during NAVA ventilation is considered like an augmented demand, and the response is a higher assistance. This concept has first been validated in healthy volunteers, with the adjustment of pressure-support amplitude in order to maintain EAdi in an optimal target during exercise (Spahija et al. 2005). Studies in adult (Brander et al. 2009a; Colombo et al. 2008), pediatric (Breatnach et al. 2010; Bengtsson and Edberg 2010), and premature patients (Beck et al. 2009) have confirmed the clinical applicability of NAVA concept. However, the clinical benefit of this concept remains to be demonstrated.

The adjustment of the ventilatory assist to the patient demand implies that this demand is continuously appropriate. Feasibility of NAVA may be impaired in particular during profound sedation. The minimal level of sedation which permits NAVA ventilation has to be studied, but individual evaluation can be done with the monitoring of EAdi during conventional ventilation. A particular concern for pediatric physicians is the immaturity of the central respiratory centers, especially during the neonatal period. Apneas are detected during NAVA, and backup ventilation is activated. However, periodic breathing risks inducing periods of low ventilation. This phenomenon remains to be evaluated, and alarms of minimal ventilation have to be set with caution.

The NAVA level is the main ventilator setting during NAVA. This proportionality factor permits to augment the ventilatory support when EAdi increases. There is a theoretic risk of excessive pressure or tidal volume. However, it has been well established in human volunteers (Beck et al. 2001), in animals (Allo et al. 2006; Beck et al. 2008; Lecomte et al. 2009), and in adult patients (Brander et al. 2009b; Colombo et al. 2008; Spahija et al. 2010) that a downregulation prevents excessive ventilation when NAVA level is elevated. When the NAVA level is titrated from zero to high level in cases of acute lung injury (Brander et al. 2009b; Lecomte et al. 2009), a two-phased response is observed (Fig. 8.31). At lower levels, airway pressure and tidal volume progressively increase, unloading the diaphragm as illustrated by a decrease of EAdi which probably return to its normal level. When the NAVA level is further augmented, the EAdi decreases further, illustrating the lower demand, and the inspiratory pressure and tidal volume will plateau. In animals, when NAVA level is at the breakpoint between the two response phases, esophageal pressure, EAdi, and PaCO₂ are similar to their baseline normal values.

The ventilatory support adjustment to the patient demand does not result in excessive ventilation. It has been shown, in premature infants (Beck et al. 2009) as in adults (Brander et al. 2009a; Colombo et al. 2008), that patients with acute lung injury ventilated with NAVA spontaneously chose a protective tidal volume (6–7 mL/kg). In pediatric patients, peak inspiratory pressures were lower during NAVA as compared to PSV, with similar minute ventilation (Breatnach et al. 2010; Bengtsson and Edberg 2010). It is however possible that very high NAVA levels induce transiently high tidal volumes or pressures, and corresponding alarms should be activated.

PEEP setting remains a challenge in intensive care. Two particularities have to be addressed with NAVA. The first point concerns the intrinsic PEEP and hyperinflation. With conventional ventilation, patients have to overcome the intrinsic PEEP before the respiratory effort can be detected. To lower this trigger delay, it is recommended to add an external PEEP. In NAVA, the trigger delay should not be altered by overdistension and this PEEP indication has probably to be reevaluated.

The second point concerns the tonic EAdi. In intubated infants, around 15 % of the inspiratory EAdi remain active at the end of expiration (Emeriaud et al. 2006). This tonic EAdi is probably involved in the control of the end-expiratory lung volume, by braking the expiratory flow. When PEEP is lowered to zero, tonic EAdi increases (Emeriaud et al. 2006; Beck et al. 2008), suggesting that tonic EAdi reflects the patient’s effort to maintain an appropriate end-expiratory lung volume. It can be hypothesized that high levels of tonic EAdi correspond to insufficient levels of PEEP. This remains to be evaluated, but tonic EAdi may become a valuable tool for the adjustment of PEEP in the future.

It has to be noticed that no data concerning periods of ventilation superior to eight hours have yet been published. The feasibility of long-term ventilation with NAVA appears good but has to be confirmed, and in particular the stability of EAdi recording.

The improvement in patient–ventilator synchrony is well established with NAVA. The clinical benefits resulting from this synchrony in terms of comfort, sleep quality, sedative use, and ventilation duration have to be evaluated with long-term ventilation periods with NAVA.

NAVA seems also to generate a “safe” ventilation with low tidal volumes. This aspect has recently been confirmed in a study with lung-injured rabbits, in which lower tidal volumes, lower airway pressures, and better PaO₂/FiO₂ were observed with NAVA as compared to protective conventional ventilation (6 mL/kg), with similar prevention of ventilator-induced lung injury markers (Brander et al. 2009a).

NAVA seems also to augment the respiratory variability (Schmidt et al. 2008). Respiratory variability is likely beneficial in ICU patients (Wysocki et al. 2006) and this effect of NAVA will have to be studied as well.