The term ''Non-invasive ventilation'' (NIV) refers to various methods of respiratory assistance, in the absence of an indwelling endotracheal tube. In recent years, the use of NIV has increased for the treatment of both acute and chronic pediatric respiratory failure. Patient tolerance to the technique is a critical factor determining its success in avoiding endotracheal intubation. One of the key factors determining tolerance to NIV is optimal synchrony between the patient's spontaneous breathing activity and the ventilator's set parameters, known as ''patient-ventilator interaction''.

Indeed, synchronization of the ventilator breath with the patient's inspiratory effort, optimizes comfort, minimizes work of breathing and reduces the need for sedation. During NIV, several factors can significantly interfere with the function of the ventilator, leading to an increased risk of asynchrony. Indeed, the presence of unintentional leaks at the patient-mask interface, the sensitivity of inspiratory and expiratory triggers, the ability to compensate for intentional and unintentional leaks and the presence/absence of expiratory valves are all factors that likely play a role in determining patient-ventilator synchronization.

The investigators therefore designed the present crossover trial in order to compare the degree of respiratory asynchronies during NIV using different ventilators (Turbine-driven ventilator vs. compressed air-driven ICU ventilators) and different setups (single circuit vs. double circuit) in children with acute respiratory failure.

After having obtained the signed informed consent from the parents of the patient, a 6 Fr pediatric esophageal balloon-catheter will be placed through a nostril in the distal third of the esophagus.

This minimally invasive procedure, will allow to monitor and record esophageal pressure swings, which are strongly correlated to pleural pressure variations and therefore allow to detect accurately patients' inspiratory efforts. Furthermore, surface electrodes will be placed in order to record the electrical activity of the diaphragm non-invasively.

In every patient, three breathing trials (30 minutes each) will be performed in randomized order:

1. NIV performed with a double limb circuit and expiratory valve incorporated in the ventilator, delivered with a pediatric/neonatal ICU ventilator (Babylog VN500, Draeger).
2. NIV performed with a single limb circuit and intentional leak (vented mask) delivered with a turbine-driven ventilator (Astral 150 [ResMed] ).
3. NIV performed with a double limb circuit and expiratory valve incorporated in the ventilator, delivered with the same turbine-driven ventilator of point 2 (Astral 150 [ResMed]).

The NIV setting decided clinically will not be modified for the study and will be held constant throughout the different study phases. Similarly, if sedative drugs are being delivered to the patient, the attending physician will decide their dose and it will be kept constant throughout the study phases. The Comfort scale will be assessed for each study phase, in order to evaluate and describe the comfort/distress of the patients during the different ventilatory strategies. Esophageal pressure tracings, inspiratory/expiratory air flows, airway pressure measured at the patient-ventilator interface and electrical activity of the diaphragm (measured with surface electrodes) will be continuously recorded with a dedicated software throughout the study in order to compute, offline, the asynchrony index (see below).

Asynchronies will be defined according to previous studies on the subject:

1. Auto-triggering (AT): a cycle delivered by the ventilator in the absence of a typical esophageal swing;
2. Ineffective Effort (IE): a deflection on the esophageal pressure monitoring not followed by an assisted cycle;
3. Late cycling (LC): a cycle with a ventilator inspiratory time greater than twice the esophageal time;
4. Premature cycling (PC): a cycle with a ventilator inspiratory time shorter than the neural inspiratory time;
5. Double triggering (DT): two ventilator-delivered cycles separated by a very short inspiratory time, during the same inspiratory Eadi signal.

The entity of asynchronies can be numerical summarized in the Asynchrony Index (AI), which is calculated as the total number of asynchrony events divided by the total number of non-triggered and triggered ventilatory cycles (expressed as percentage).

Asynchrony Index (%) = [(AT + IE + LC + PC + DT) / (RRpes + AT)]×100 Where AT refers to Auto-triggering, IE to ineffective triggering, LC to late cycling, PC to premature cycling, DT to double triggering and RRpes to the respiratory rate as measured using the esophageal pressure tracing.

Furthermore, the number of each type of asynchrony will be assessed (number of events per minute), in order to identify the most relevant types of asynchronies.

Randomization The randomization of the three NIV-phases will be performed with an online randomization software called "Research Randomizer" (https://www.randomizer.org). No risk of bias is foreseen, as all patients will undergo the three interventions (cross-over study).

Blinding. The respiratory traces registered during the different study phases and analyzed offline in order compute the "Asynchrony Index" will be evaluated by an investigator blinded to the type of intervention.

PRIMARY ENDPOINT Primary endpoint of the present study is the difference in Asynchrony Index (expressed as %) obtained during NIV performed with an ICU ventilator using a double limb circuit and the value obtained during NIV performed with single limb circuit with intentional leak with a turbine-driven ventilator.

Secondary endpoint Secondary endpoint of the present study is the difference in Asynchrony Index (expressed as %) obtained during NIV performed with an ICU ventilator using a double limb circuit and the value obtained with the same type of circuit, but with a turbine-driven ventilator.

STATISTICAL ANALYSIS Sample size calculation. The sample size for the primary endpoint of the study has been calculated using the software G*Power 3.1.9.2 using a paired t-test and using as outcome parameter the difference in Asynchrony Index (AI) during NIV performed with ICU ventilators and with turbine-driven ventilators applied with single limb circuit and intentional leaks. Based on available data the investigators estimated in our population an AI of 59±13% and considered a 20% reduction of its value as clinically relevant (AI=47±13%). Considering a two-tailed alfa error of 0.05 and a desired power of 0.8, with an effect size of 0.923 the investigators calculated a sample size of 12 patients.

DATA ANALYSIS All data will be tested for homogeneity of variance and normality of distribution using the Shapiro- Wilk test. Normally distributed data will be expressed as mean ± standard deviation, while nonnormally distributed data as median and interquartile range. The presence of outliers will be carefully assessed during evaluation of distribution of data; however, no action is foreseen to exclude outliers.

Variables (Asynchrony Index, respiratory rate, tidal volume, minute ventilation, esophageal pressure variation, etc.) recorded during the different NIV modalities will be compared via paired t-test or Signed Rank Sum test, as appropriate. Mean difference and its 95% CI will be calculated for normally distributed data. For non-normally distributed variables, median difference and its 95% CI will be estimated by Hodges-Lehmann's median analysis. All tests will be two tailed and statistical significance is defined as p<0.050. Analysis will be performed with SigmaPlot v.12.0 (Systat Software Inc., San Jose, CA) and SAS 9.2 (SAS Institute Inc., Cary, NC, USA).

Of note, the above-noted statistical procedures are appropriate but will not exclude other procedures that may also be used in addition to or in lieu of the stated procedures in order to best analyze the data. No control subjects will be needed, as each patient will serve as its own control for the subsequent measurements (cross-over study).

Inclusion Criteria:

• Patients with acute hypoxic (SpO2/FIO2 ratio < 315) or hypercapnic (PvCO2 > 52 mmHg and venous pH <7.28) respiratory failure in which non-invasive respiratory support is clinically indicated
• Age: > 28 days and < 4 years
• Patients whose parents provided signed informed consent

Exclusion Criteria:

• Age > 4 years or < 28 days
• Patients whose parents did not provide signed informed consent
• Clinical contraindications to non-invasive ventilation
• Clinical contraindication to the placement of an esophageal balloon

Analysis of respiratory tracings will be blinded to the type of respiratory support.