Patient-ventilator synchrony in Neurally Adjusted Ventilatory Assist (NAVA) and Pressure Support Ventilation (PSV): a prospective observational study

Our study showed that a lower total number of asynchronies in NAVA compared with PSV
reflected an improved patient-ventilator interaction. This finding is consistent with
results of previous studies 10], 11], 13]–15]. This reduction in the number of asynchronies is related to a reduction in ineffective
efforts and auto-triggering. Ineffective efforts could be due to the presence of intrinsic
PEEP, which increases the patient’s effort required to trigger the ventilator. Ineffective
efforts could also be due to excessive levels of pressure. The consequences of ineffective
efforts are that the patient’s inspiratory effort will fail to trigger a ventilator
breath 9]. In our study, we attempted to optimize the pressure level and the expiratory trigger
in the PSV mode. Unlike other studies 10], 11], 13], 15], our study showed persistence of ineffective efforts in NAVA. This finding could
be related to the fact that the trigger in NAVA relies on the principle “first detected,
first served”.

The ventilator initiates the machine cycle according to the first trigger it has detected.
This trigger is a pneumatic trigger or a trigger based on the EAdi signal. In our
study including 56.6 % of patients with a known respiratory disease, we observed that
some patients used their accessory breathing muscles to trigger the ventilator. Therefore,
a pneumatic trigger was rewarded before onset of the diaphragmatic signal. The machine
cycle beginning before the EAdi signal, and having no way to detect if the data stored
is a pneumatic trigger or trigger signal based on the Eadi. We could analyze some
cycles triggered by a pneumatic trigger as asynchronous cycles (auto-triggering followed
by ineffective efforts). Another possible explanation for this auto-triggering is
that we took into account all of the diaphragmatic signals in our analysis. However,
some signals may correspond to artifacts (cardiac activity). Mauri et al. 13] excluded some diaphragmatic signals, considering them as artifacts, in their study.

In our study, double triggering was more frequent in NAVA than in PSV. This finding
is consistent with the results of Piquilloud and colleagues 10]. This larger number of double triggering in NAVA is related to the fact that sometimes
there are EAdi signals with a biphasic appearance, and this causes two successive
cycles (Fig. 3). This biphasic appearance could be related to early cycling when the inspiratory
time of the ventilator is less than the neural inspiratory time of the patient. This
may not increase the work of breathing, but it may participate in the discomfort felt
10].

In our study, the NAVA level was not optimized, which could have affected the persistence
of asynchronies. The initial setting of the NAVA gain was based on the level of assistance
in PSV. In most other studies 10], 11], 14], 15], the NAVA gain was set in the same way as in our study using the “NAVA preview” function
of the ventilator Servo-I, and this technique is recommended by the manufacturer.
In addition, the NAVA gain was not changed much during the nychthemeron, unlike the
level of assistance in PSV. This finding is probably due to less control of this new
type of ventilation compared with PSV, which is the reference mode.

A better way to settle the NAVA level might be by using the method of Roze and colleagues
17]. Their method is based on daily titration of the NAVA level according to the maximum
EAdi signal that is obtained during a spontaneous breathing trial with a level of
support of 7 cmH
₂
O and a PEEP level of 0 cmH
₂
O. We unexpectedly found a relatively low AI in NAVA and PSV. This finding can be
explained by the fact that all of our settings, at least in PSV, were optimized with
particular attention to setting the level of assistance and expiratory cycling depending
on the condition of the patients.

To the best of our knowledge, this is the first study to compare asynchronies in PSV
and NAVA where ventilator settings in PSV have been optimized. However, we did not
analyze other types of patient-ventilator asynchronies, such as early or late cycling,
which could artificially underestimate the total number of asynchronies, and thus
the AI. Thille and colleagues 6] showed that an AI greater than 10 % is associated with an increase in the duration
of mechanical ventilation and an increase in use of tracheotomy for ventilator weaning.
In our study, less patients using NAVA than those using PSV had an asynchrony index
greater than 10 %. NAVA could be helpful in patients with difficult weaning by reducing
the number of asynchronies, particularly in patients with a high AI.

We did not find any difference in VT between NAVA and PSV, similar to most previous
studies 10], 13], 15]. However, with NAVA, more patients had a VT between 6 and 8 ml/kg of PBW than those
with PSV. This finding suggests that in NAVA, over 23 h, periods of over- and under-assistance
are relatively limited compared with PSV.

This setting of VT between 6 and 8 ml/kg is recommended in protective ventilation
in acute respiratory distress syndrome to reduce the risk of barotrauma and volutrauma
18]. However, some studies appear to suggest that this protective ventilation may also
reduce the risk of ventilator-induced lung injury in patients without acute respiratory
distress syndrome and those who are ventilated invasively 19], 20]. Several studies have demonstrated a reduced risk of over-assistance in NAVA compared
with PSV 14], 15], 21]. In contrast, few studies have focused on the risk of under-assistance.

In a previous study of postoperative patients with thoracic or abdominal surgery,
a few patients had a VT less than 5 ml/kg, with no signs of discomfort or respiratory
distress 16]. However, the setting of VT is normally between 6 and 8 ml/kg of PBW, and a lower
level of VT should not be used only on the basis of this previous study. However,
we can assume that for some patients with good clinical and biological tolerance,
a lower VT can be accepted. In NAVA, the majority of previous studies found variability
in VT, and this is one of the benefits, at least theoretically, of proportional ventilation
modes 12], 15], 16], 22]. In our study, we do not find this variability in VT. This lack of finding is probably
due to the small size of our population. In addition, when recording VT on paper,
at a given moment, this variability might be hidden. However, we also examined the
number of patients who had variation in VT of more than 13 %, which corresponds to
the median of the variability of VT in our series. We found that more patients in
NAVA had variability of VT greater than the median of variability than those in PSV.

In contrast to Piquilloud et al’s study 10], but in agreement with Terzi et al’s study 9], we found improvement in the parameters of oxygenation in NAVA. These findings could
have resulted from better patient-ventilator synchronization and a more natural breathing
pattern, which may also contribute to improved gas exchange 9].

Our study has some limitations. First, the absence of randomization could have introduced
bias in the study in terms of adaptation to the ventilatory mode, but the study aim
was not to compare the duration of weaning. Second, not analyzing early or late cycling
could have led to underestimation of the total number of asynchronies, and thus the
AI. Early or late cycling is observed with the PSV mode, but not with the NAVA mode
10]. Despite recording for 23 h, analysis of respiratory curves was manually analyzed
for the first 5 min of recording every 4 h, which represents an analysis period of
25 min.

Finally, our study, which recorded data for 23 h, similar to Coisel and colleagues’
study 16], confirms the stability of the EAdi signal. This appears to be the primum movens
before considering this new ventilation mode as a potential mode of weaning from mechanical
ventilation.