Mechanical ventilation permits to support the work of breathing in case of respiratory failure, but therapy also has many side effects. Non-invasive ventilation (NIV), which delivers the ventilatory assist via a face mask or nasal canula, permits to decrease these complications. However, NIV is not always successful and half of children in respiratory failure finally require invasive ventilation. A major cause of NIV failure is the ventilator inability to detect patient efforts. The new ventilatory mode NAVA (neurally adjusted ventilatory assist) improves the detection of patient efforts during mechanical ventilation. The hypothesis of this study is that NAVA improves synchrony during pediatric NIV and therefore permits to unload the patient ventilatory efforts.

Nasogastric tube installation to monitor electrical activation of the diaphragm (EAdi):

A specific nasogastric tube equipped with an array of microelectrodes (Edi catheter, Maquet, Solna, Sweden) will be installed after inclusion. This tube has an external aspect and size (6F, 8F, or 12F depending on the patient size) similar to usual nasogastric tubes. The gastric tube installation is extremely frequent in pediatric ICU, and systematic in patients with ventilatory support, to empty the gastric gas and permit the feeding. The new tube will stay in place after the study and could be used as a classic tube. The catheter position in esophagus will be adjusted using a special window of the Servo i ventilator screen, which permits to confirm the correct positioning close to the diaphragm [Barwing 2009]. It has recently been confirmed that Edi catheter placement is not difficult and that the correct placement of the nasogastric tube is actually facilitated by the possibility of activity monitoring [Green 2011].

RIP jacket installation:

A special jacket adapted to the patient size will be installed to monitor the lung volume changes by impedance plethysmography (RIP). This sleeveless jacket is constituted of a large band of distensible cloth in which 2 metallic coils are coated to record the impedance changes created by the ventilation. The RIP coils will be connected to the RIP monitoring module, and no volume calibration will be done to avoid the manipulation of patient airway. RIP monitoring is completely non-invasive and no complications are expected [Emeriaud 2010][Emeriaud 2008].

Baseline measurements:

The Edi tube is connected to the Servo i to monitor EAdi, but the patient remains connected to its usual ventilator with unmodified ventilator settings. Airway pressure, respiratory volume and flow will be recorded simultaneously with EAdi during 30 minutes.

Ventilator change:

If a different ventilator was used prior to the study, the patient will be installed on a Servo i ventilator, initially with the same ventilatory settings.

Determination of NAVA parameters:

The positive end expiratory pressure (PEEP) will not be modified. The NAVA level - the proportionality factor that converts EAdi (microV) into pressure support (cmH2O) - will be set using a specific window which simulates the pressure that would be delivered in NAVA. The NAVA level will be initially adjusted to match the simulated pressure with the actual assist pressure. If the patient is on CPAP (i.e. no assist pressure), the NAVA level will be initially set to match an assist pressure of 5 cmH2O above PEEP. The maximal pressure alarm will be set at 30 cmH2O.

NAVA period:

The NAVA mode will be activated for one hour, under the continuous supervision by a physician involved in the study, and with a continuous monitoring of vital signs including cardiac and respiratory rate, SaO2, respiratory distress signs, patient agitation, and the normal function of the ventilator. The NAVA level will be adjusted if the patient breathing frequency is > 40/min (progressive increase of NAVA level) or < 12/min (decrease of NAVA level), or if the delivered pressure is low (<3 cmH2O) due to air leaks. Ventilatory pressure, flow and volumes and EAdi will be continuously recorded during the last 30 minutes.

Second period with conventional NIV settings:

The patient will be installed again with the ventilatory settings prescribed prior to the study, and the ventilatory data will be recorded again for 30 minutes.

During the entire study, "usual" modifications of settings (e.g. adaptation of assist level, FiO2, or PEEP) considered by the attending physician, the nurse, or a respiratory therapist, will be permitted and documented.

End of the study:

At the end of the study, if a clear clinical benefit of one mode or setting is observed during the study, the attending physician will be informed to permit the optimal adaptation of the ventilatory assist.

Inclusion Criteria:

• Children >3 days and <18 years old
• Hospitalized in the pediatric intensive care unit, and requiring non invasive ventilation (including any ventilatory modes, with one (CPAP) or two level of pressure assist) for more than 6 hours according to the prescription of the attending physician

Exclusion Criteria:

• Contraindications to NAVA use or to the placement of a new nasogastric tube (e.g. bilateral phrenic paralysis, trauma or recent surgery in cervical, nasopharyngeal or esophageal regions)
• Hemodynamic instability requiring dopamine ≥ 5µg/kg/min, epinephrine, norepinephrine, or dobutamine.
• Severe respiratory instability requiring imminent intubation according to the attending physician, or FiO2 > 60%, or PaCO2>80 mmHg on blood gas in the last hour.
• Absence of parental or tutor consent.
• Patient for whom a limitation of life support treatments is discussed or decided