Chronic obstructive lung disease is a disabling disease that affects people usually after several years of smoke tobacco exposure and affects millions of patients worldwide. The disease is marked by multiples episode of worsening, termed exacerbations necessitating frequent hospitalizations. During these exacerbations, patients present breathless, and in the most severe cases, are admitted to an Intensive Care Unit (ICU) for respiratory assistance. Currently, respiratory assistance is provided by a ventilator via a oronasal mask (referred to non-invasive ventilation, NIV), that helps patients to cope with their breathless. The mask is not always well tolerated and the ventilator sessions are delivered intermittently. In the past decade, a new technique that provides air-oxygen with high flow has been developed. This technique, called High Flow via Nasal Cannula (HFNC) can deliver from 21 to 100% heated and humidified air-oxygen at a high flow of gas via simple nasal cannula. Recent studies have shown that the technique is very efficient to treat patients presenting with acute respiratory failure who don't have any underlying chronic pulmonary disease. Whether the technique would be also efficient in patients with COLD presenting with severe exacerbations has not yet been demonstrated. Since HFNC does not require any mask, it is thought that the comfort of the patient would be much better in comparison to NIV and could potentially help to treat many patients with the disease. The objective of the present study is to study the physiological effect of HFNC as compared to NIV in patients with severe exacerbations of COPD and to show that it is non-inferior to NIV.

Design: Prospective, observational, cross-over, clinical physiologic study of HFNC and NIV in patients with COPD and hypercapnic respiratory failure Intervention: Participants will be consecutively ventilated with the two devices, i.e., HFNC and NIV. HFNC 30 and 50 L/min will be applied in a random order (sealed, opaque envelope). HFNC will be set with a temperature at 37°C or 34°C if perceived as too warm. NIV settings will be adjusted based on the clinical assessment of the Respiratory Therapist as per standard practice and will not be modified during the test. In both groups, FiO2 will be adjusted to achieve a SpO2 of at least 92%.

Setting: The intervention will be applied in the Medical-Surgical ICU of St. Michael's Hospital.

Duration: Each device will be studied for up to 30 minutes. The entire study will take place over up to 3 hours. There is no additional follow-up thereafter.

Procedures: The study will start first by recording patients under spontaneous breathing considered as a baseline (sequence #1). Under spontaneous breathing, patients will be receiving oxygen therapy as it is usually done between 2 NIV sessions. Then, the patient will have up to 30 minutes to acclimatize to HFNC and select his/her preferred flow rate (sequence #2). NIV will be applied (sequence #3). The NIV settings will be adjusted based on the clinical assessment of the Respiratory Therapist and will not be modified throughout the study. Then, according to randomization order (sealed envelopes), patients will receive a flow corresponding to the preferred flow selected by the patient and a second flow, either HFNC 30 L/min or HFNC 50 L/min depending on the initial choice (sequences #4 and #5) (see flow diagram). FiO2 will be adjusted to achieve a SpO2 of 90 to 94%. Each condition will be studied fup to 30 minutes and will be separated by a 5-minute washout period during which patients will be receiving oxygen therapy. The measurements will be collected at baseline (under spontaneous breathing) and during the last five minutes of each condition. In case of clinical intolerance as considered by the attending physician, the study will stop. In case the patient is considered dependent of NIV by clinicians, we will still enroll the patient without doing baseline period after discussion with clinicians.

The pulse oximeter (SpO2) and transcutaneous CO2 monitor (SenTec Digital Monitoring System (SDMS)) will be continuously monitored throughout the study period. Exspiron or Electrical Impedance Tomography (EIT) device will be calibrated to measure minute ventilation. Four surface electrodes will be placed bilaterally to record each hemidiaphragm activity using a specific recording system (Acqknowledge software, Biopac Systems) A bedside ultrasound examination will be performed, using a SonoSite system (Fujifilm) equipped with a 10-15 MHz ultrasound linear probe. Diaphragm thickness will be measured both at end-inspiration and end-expiration. This technique will be applied at the end of each sequence (before ventilatory assist under conventional O2 therapy). In patients becoming rapidly intolerant to NIV disconnection, we will limit this "baseline" period under O2 to a minimum (5 minutes). The thickness and contraction of the intercostal muscles and abdominal muscles will also be assessed.

Interpretation of the ultrasound results to determine diaphragm and other muscle thickness will be read at a later time by an individual blinded to the intervention.

Inclusion Criteria:

1. COPD exacerbation and acute hypercapnic respiratory failure with acute respiratory failure defined by

   • Respiratory acidosis (pH ≤7.35 and PaCO2 ≥45 mmHg);
   • Respiratory rate≥20 breaths/min;
   • Activation of accessory respiratory muscles;

2. Undergone at NIV or HFNC since their admission

3. English speaking

4. Adult patient with age > 40 year old.

Exclusion Criteria:

1. Severe respiratory acidosis defined by pH<7.25
2. Decreased level of consciousness (Glasgow Coma Score Scale < 11)
3. Urgent intubation required
4. Pneumothorax with pleural drainage and persistent air leak
5. Hemodynamic instability requiring vasopressors
6. Uncooperative
7. Patients with skin or chest wall or abdominal trauma (potentially worsened by placement of a surface sensor)
8. Clinical judgement of the attending physician
9. Body mass index > 40 kg/m2