Comparison of High Flow Nasal Cannula (HFNC), Face-mask Non-Invasive Ventilation (NIV) & Helmet NIV in COVID-19 ARDS Patients

Objective: To determine whether NIV delivered through helmet interface reduces intubation rate among patients with COVID-19 ARDS compared to face-mask NIV and HFNC. Design, setting \& participants: Two-center randomized clinical trial of 360 patients with mild to moderate ARDS and confirmed COVID-19 requiring non-invasive ventilation between August 2020 to January 2021. The patients with respiratory rate (RR) more than 30/min or oxygen saturation (SpO2) less than 90% or PaO2/FiO2 ratio less than 300 despite standard oxygen therapy by face mask (\<15 L/min) who present to Royal hospital or Sultan Qaboos University Hospital (SQUH) emergency department, medical wards or intensive care unit (ICU). Intervention: Patients will be randomly assigned (block randomization) to either face-mask NIV, HFNC or Helmet NIV. The helmet is a transparent hood that covers the entire head of the patient and has a rubber collar neck seal. Main outcome and measures: The primary outcome is the rate of endotracheal intubation at 28-days. Secondary outcomes include hospital mortality at 28 and 90 days, NIV free days, invasive ventilator free days and hospital length of stay. Expected results: We assume the failure rate of Helmet NIV to be 30%, failure rate of HFNC to be 40% and failure rate of face-mask NIV to be 50%. A sample size of 360 patients (120/group) will achieve a power of 0.90 at a significance level of 0.05. To account for 10% dropout rate, the total sample required is 396 subjects(132/group).

Inclusion criteria: * \> 18 years of age * confirmed COVID-19 * Within 48 hours of presentation in the emergency department, high dependency area or intensive care unit (ICU) * ARDS according to Berlin definition (P/F \< 300) or O2 saturation \< 90% or RR \> 30/min) in room air * Standard oxygen therapy at flow rate \< 15L/min x 60 minutes Exclusion criteria: * More than 48 hours of admission to the emergency room, high dependency or intensive care unit * On non-invasive ventilation or HFNC for more than 4 hours before enrollement * Impending cardiopulmonary arrest * Need for immediate endotracheal intubation * Hemodynamic instability or life-threatening arrhythmias * GCS \< 8 * Active intracranial pathology or high ICP * Inability to cooperate or protect the airway * Pregnancy * Tracheostomy * DNR or refusing intubation * Known type 2 respiratory failure * On chronic home oxygen therapy Enrolment: Multi-centre randomized trial of confirmed COVID-19 cases presenting to the emergency department, the ward, high dependency or intensive care unit (ICU) with ARDS requiring non-invasive mechanical ventilation. It is an open label randomized clinical trial. We will perform permuted block randomization of 9 and perform concealment. Consent: A written consent must be taken from the patient or the decision maker prior to enrollment. Each of the three arms have pre-specified treatment protocol in the form of algorithm. For the helmet NIV group, for type 1 respiratory failure (hypoxemia \& normal pCO2), CPAP will be started at 10 cmH2O \& FiO2 0.6. Re-assessment will be done every 15-30 minutes and CPAP and FiO2 will be increased accordinly (CPAP to 15 \& FiO2 of 1.0 maximum) If patients develops hypercapnia, the mode will be changed to BiPAP: IPAP 12-25 cmH2O (max 20) EPAP 8-15 cmH2O. In the face-mask NIV group, CPAP will be started at 5 cmH2O \& FiO2 0.6. Re-assess will be performed every15-30 minutes. Thereafter CPAP \& FiO2 will be adjusted accordingly (maximum CPAP 12 \& FiO2: 1.0) If the patient develops high pCO2 :BiPAP: IPAP 8-20 cmH2O (max 20) EPAP 4-12 cmH2O. In the high flow nasal cannula group, HFNC will be started at at 20-30 L/min Adjust FiO2 to keep SpO2 \> 92%. Thereafter, it will be titrated to flow to max 60 L/min \& FiO2 100% For all the three groups, these are the following targets: Target: SPO2 ≥ 92 % RR ≤ 30/min ROX index \> 2.85 at 2 hours and \> 3.85 at 12 hours (HFNC group) If the above targets are not achieved, the ICU team will assess them for possible endotracheal intubation. In case the patient doesn't tolerate any of the two NIV interfaces, they will be allowed to cross over to the other with intention to treat analysis. For the high flow oxygen group, a trial of non-invasive ventilation will be allowed at the physician discretion before endotracheal intubation with intention to treat analysis. Statistical Analysis: We tested the null hypothesis that there is no difference between the Helmet-NIV, HFNC and face-mask-NIV intubation rate and the alternative hypothesis that Helmet NIV and HFNC is superior to face-mask-NIV. We assumed the failure rate of Helmet NIV to be 30% and the failure rate of HFNC to be 40% and failure rate of face-mask NIV is 50%. The group subject counts are allocated 1:1:1. The effect size is 0.12. A sample of 360 subjects, (120/group) achieves a power of 0.90 at a significance level of 0.05. To account for 10% dropout rate, the total sample required is 396 subjects(132/group). Baseline characteristics will be reported by mean and standard deviation for the continuous variables and numbers and proportions for categorical variables. Intention to treat analysis to analysis will be used for the results for the primary efficacy point of reduction in the rate of the intubation between the three groups. Fisher's exact test will be used between-group differences in the rate of intubation. The face-mask-NIV is considered as a control group as it is the standard treatment and will calculate t the relative risk and absolute risk difference for the primary end points. The log-rank test will compare the time from enrollment to ET intubation for the three groups using Kaplan-Meier curves plot. Similar analysis will be performed for the 28-day mortality for as the secondary end points. All P values reported for on primary and secondary end points are based on two-sided tests of 0.05. The SAS statistical software is used to analyze the data.