S Boet and colleagues
Multicentre RCT of HBOT for COVID-19
Research Proposal

Intubation and mechanical ventilation are associated with >50% mortality in COVID-19 patients.15 These modalities also overwhelm limited intensive care unit (ICU) resources, and may lead to further mortality if patients do not have access to care. The pathophysiology of severe COVID-19 pneumonia includes decreased oxygen diffusion from the alveoli to the blood, and a massive pro-inflammatory cytokine response.6 HBOT consists of breathing 100% oxygen at a pressure >1 Atmosphere Absolute (ATA). This increases tissue oxygen delivery 10-20-fold and also has strong anti-inflammatory effects.7,8 Case series from China, France and the United States show reduced ICU admission and intubation rates and shortened hospital stays when hypoxemic COVID-19 patients are treated with HBOT.912 Although promising, these results require a well-conducted multicentre randomized controlled trial (RCT) to determine the effectiveness of HBOT for improving oxygenation, morbidity, and mortality among hypoxemic COVID-19 patients. Our interdisciplinary team has expertise in hyperbaric medicine, critical care, and multicentre randomized controlled trials (RCTs). We propose a multicentre RCT involving hypoxemic hospitalized COVID-19 patients to test the hypotheses that HBOT significantly improves clinical outcomes by reducing the duration of oxygen, rates of intubation and ICU admission, length of stay and mortality. We hypothesize that HBOT will improve clinical outcomes in COVID-19 patients and ultimately save lives.
1.1. What is the problem to be addressed?
1.1.1 Mortality is high among COVID-19 patients admitted for oxygen supplementation: Approximately 15 to 20% of COVID-19 patients present with hypoxemic respiratory failure requiring oxygen supplementation.13 More than one third of patients that require low flow oxygen on presentation, and 60% of those that require higher flow oxygen, will require intubation and mechanical ventilation.1,5,14 Intubation and mechanical ventilation in themselves are associated with mortality rates between 50% and 97% in COVID-19 patients.
1.1.2 Mechanical ventilation is an invasive and limited resource: The current SARS-CoV-2 viral pandemic has infected over 3.6 million individuals worldwide and over 70,000 people in Canada, with over 5,000 deaths as of May 12, 2020.15 The current pandemic has a huge potential burden on the healthcare system. Interventions such as mechanical ventilation and extracorporeal membrane oxygenation (ECMO) are limited resources. Government efforts to flatten the curve are being made to keep the number of patients requiring critical care within the available resources. Therefore, any intervention that prevents further clinical deterioration and associated invasive treatments for COVID- 19 patients may help to sustain the healthcare system.
1.1.3 Hypoxemia and “cytokine storms are crucial pathophysiological factors leading to death: The pathophysiology of severe COVID-19 combines a decreased oxygen diffusion from the alveoli to the blood as well as an overwhelming inflammatory response referred to as the ctokine storms. 16,17 Critically ill COVID-19 patients have presented with high blood concentration of pro-inflammatory cytokines including IL2, IL7, GCSF, IP10, MCP1, MIP1A and TNF.18 Higher concentrations of IL-6 at admission are independently associated with in-hospital mortality.14
1.1.4 Day 5 to Day 9 is a unique window of opportunity to intervene: The most common reason for hospital admission is hypoxia requiring supplemental oxygen. Several convergent studies describe a pivotal period, between Days 5 and 9 from disease onset, when patients with COVID-19 frequently require admission for oxygen supplementation due to worsening respiratory distress and hypoxia.5,18,19 Progression to acute respiratory distress syndrome (ARDS) typically occurs after day 8.5 During that time period, early HBOT for hypoxemic COVID-19 patients could correct the ongoing tissue hypoxia and limit the cytokines storm, preventing further clinical deterioration and the need for mechanical ventilation.
1.1.5 What are the desirable characteristics of a treatment intervention for severe COVID-19 patients? An appealing treatment intervention would be: (i) an already approved drug/intervention for non-COVID-19 indications that is (ii) minimally invasive, (iii) safe to patients and healthcare professionals, (iv) less expensive than current standard of care, (v) targeted at patients who are the most at risk of severe complications and death, and (vi) effective on clinically meaningful and patient centered outcomes such as decreasing intubation and mechanical ventilation, critical care admission, length of stay and mortality. HBOT meets each of these criteria.
1.2. Hyperbaric oxygen therapy (HBOT) may be an important intervention to consider
1.2.1 What is HBOT? HBOT for clinical use is defined as breathing 100% oxygen pressures >1.4 ATA in a specialized chamber (Figure 1).7 HBOT is a well-established and safe20 method to increase tissue oxygen delivery up to 10 fold at 2 ATA pressure.8,21,22 Hyper-oxygenation of arterial blood with plasma- dissolved oxygen has a strong anti-inflammatory2225 effect and may have a direct viricidal effect on SARS-CoV-2 HBOT is currently approved by Health Canada for 14 indications for both elective (e.g. soft tissue radiation therapy complications, non-healing chronic wounds) and urgent conditions (e.g. carbon monoxide poisoning, decompression sickness, gas
1.2.2 A sound physiological and preclinical rationale is
compelling to suggest an effectiveness of HBOT for severe COVID-
19 patients through three mechanisms:

Increased oxygen delivery (Henry's law): HBOT promotes
oxygen transport, by significantly increasing dissolved oxygen
concentrations in the blood and delivery to the tissues.8,21,22 In severe
COVID-19 patients, the barrier to diffusion is in the alveoli
(inflammatory exudate-pneumonia) and inflamed interstitium.
Standard oxygen therapy at ambient pressure increases the partial
pressure of oxygen in the alveoli by increasing the inspired fraction of oxygen (FIO2). However, as the pneumonitis progresses, high FIO2 at ambient pressure cannot penetrate the diffusion barriers in the lungs. HBOT overcomes diffusion barriers by further increasing the partial pressure of oxygen.
Anti-inflammatory: A large body of evidence both from preclinical and clinical studies, including from our group, shows that HBOT has an immunomodulatory effect regulating the inflammatory response.2225 Both the humoral and cellular immune response are stimulated by HBOT. HBOT stimulates the activities of neutrophils.26 Through complex physiological cascades involving reactive oxygen species (ROS), HBOT decreases pro-inflammatory cytokines including IL-1, IL-6, IL-18, TNF, and increases anti-inflammatory cytokines.22,24,25 Intervening early to limit the increase of plasma IL-6 may be beneficial to these patients, in particular since elevated levels of IL-6 are independently associated with mortality for COVID-19 patients.14
 A suspected direct viricidal action that destroys the envelope of the virus: A direct viricidal action has been demonstrated in preclinical research in non-SARS-CoV-2 enveloped viruses.27 Studies on the human immunodeficiency virus us (HIV) have shown that HBOT, through ROS, has a viricidal effect on enveloped viruses.27 HBOT upregulates the hypoxia inducible factor (HIF) which promotes the expression of human antiviral peptides, such as defensins and cathelicidins, effective to block the coated, positive-sense single-stranded RNA virus (such as SARS-CoV-2).28,29
1.2.3 Promising initial case reports and series of HBOT in severe COVID-19 patients: Chinese clinicians reported a case of a 69-year old patient presenting signs of respiratory decompensation. The expected course of the disease reversed after repeated sessions of HBOT and mechanical ventilation was not required.10 Five other similar cases have been reported from China.9 In Louisiana, a series of 11 severe COVID-19 patients who required oxygen supplementation were successfully treated; none of them required intubation.11 In France, an ongoing RCT investigating the effectiveness of HBOT on COVID-19 patients requiring oxygen supplementation12 has enrolled 18 patients as of May 5. One patient from the control group (n=9; conventional treatment with no HBOT) has been admitted to ICU versus none from the HBOT group (n=9). Patients from the HBOT group were weaned off oxygen and discharged from the hospital faster (median length of stay of 5 days with HBOT versus 10 days for standard of care).

1.2.4 HBOT is safe and inexpensive: HBOT is a non-invasive and low-risk intervention when contraindications are respected, including for lungs.30,31 HBOT is also very low risk for healthcare professionals who supervise the treatment. In our centers, rigorous and systematic safety measures are in place. Patients are monitored from outside the chambers, and extra infection control measures for COVID-19 have been added. HBOT costs between C$120 and C$500 per treatment depending on the type of chambers and personnel costs.32,33 This is significantly less expensive than the daily costs of mechanical ventilation, which is over $3,500 daily in Canada.34 The relatively low cost of providing HBOT along with its potential to improve the prognosis of severe COVID-19 patients make this intervention worth studying, despite the current limited number of HBOT centres. Evidence of safety, effectiveness, and cost savings will encourage governments to increase access to HBOT in the immediate future and reduce the strain of COVID-19 on the healthcare system. This could be feasibly and quickly accomplished given the availability of medical portable chambers (up to 3.0 ATA) and the short time needed to train personnel (within weeks). Increasing access to HBOT may also benefit more patients beyond the current pandemic for already approved indications.

1.3. How will the results of the trial be used? If HBOT improves outcome and prevents further deterioration leading to critical care for severe COVID-19 patients, practice will change internationally. If no benefit is found from the intervention, then the current standard of care (no HBOT) will be supported by level I evidence.

1.4. What are the hypotheses to be addressed in the HBOT for COVID-19 Trial?
We hypothesize that when daily HBOT is administered to hospitalized COVID-19 patients on oxygen supplementation compared to standard care: (a) clinical status at day 7 will improve; and (b) time to wean from oxygen, rate of mechanical ventilation, rate of ICU admission, length of stay in ICU and in hospital, overall mortality at 28 days, biological inflammation, and costs of care will be reduced.

2.1. Proposed trial design: This Canadian-led international trial is a sequential Bayesian Parallel-group, individually Randomized, Open, Blinded Endpoint (PROBE) controlled trial at five centers across 3 countries (Ottawa, Toronto, Edmonton, Rugby/Coventry (UK), Geneva (Switzerland). We are using a design approach and statistical analysis plan developed by Dr. Frank Harrell and made publicly available for COVID researchers.27 Fast-tracked researchers approval of the common study protocol will be obtained from the local ethics boards and from Health Canada for use of HBOT in COVID-19 pneumonia.

2.2.1. Planned intervention: The scheduled HBOT treatment will take place in existing hospital-based hyperbaric chambers (multiplace or monoplace). In the air pressurized multiplace chamber, the patient will sit and breathe through a hood. In the oxygen pressurized monoplace chamber, the patient will be placed in a semi-recumbent position without the need for a hood. The sessions will be supervised by a hyperbaric physician and chamber controller located outside the chamber. The HBOT session will start with compression over approximately 10 minutes while breathing oxygen, until a pressure of 2.0 ATA is reached (equivalent to 10 metres of sea water). The maximum oxygen partial pressure inhaled by the patient will be 1,500 mmHg (Figure 2). This will be followed by breathing 100% FIO2 inhalation for 25 minutes at 2.0 ATA followed by breathing air (FIO2=21%) for five minutes at 2.0 ATA. There will then be a second 25-minute period breathing 100% FIO2 at 2.0 ATA, followed by decompression under 100% FIO2 at 1 meter/min to ambient pressure (1.0 ATA). The total duration of the session will be 75 minutes, given once a day. HBOT sessions are typically conducted at 2.0-3.0 ATA for 60-90 min, once a day. We chose the dose of HBOT (2.0 ATA, 75 min, 1 session/day) to balance risks, effectiveness, and logistics. Before the first HBOT session, the hyperbaric operator will educate the patient on wearing of the hood and middle ear pressure equalization maneuvers. Before each HBOT session, each patient will have their vitals recorded, contraindications reviewed, and a safety checklist completed. The risks are those inherent in a standard HBOT. The hyperbaric units included in this trial are all experienced and conduct several thousands of HBOT annually. Compression and decompression speeds are slow to minimize the risk of barotrauma. The hyperoxic risks of oxygen toxicity are low and controlled in this context of hypoxemia as: (1) the risk of developing a hyperoxic crisis is at most 1/6000 sessions;20 and (2) the cumulative dose of oxygen linked to daily HBOT sessions remains low, in the order of 200 UPTD (Unit of Pulmonary Toxic Dose). To further decrease any theoretical pulmonary or neurological risk of toxicity, we will limit the pressure at 2.0 ATA and will add a 5-minute air break. Breathing air at 2.0 ATA ensures maintenance of normal oxygen saturation, even in patients that would be mildly to moderately hypoxemic in normobaric room air. Between their HBOT sessions, patients from the intervention group will receive usual care, including normobaric oxygen supplementation as per standard of care.

2.2.2. Control group: Current standard of care treatment, including normobaric oxygen.
2.3. Allocation & randomization: The random allocation sequence will be computer-generated by an independent statistician using permuted blocks of randomly varying lengths, stratified by center, biological sex and age (<60; 60). COVID-19 epidemiology and prognosis vary by biological sex and age.15 Study personnel will access the randomization sequence via a central web-based application to ensure allocation concealment.
2.4. Protection against bias: Minimal exclusion criteria to allow a diverse population will enhance external validity. Outcome assessors will remain blinded to reduce performance bias.
2.5. Duration of treatment: Daily HBOT will be administered until either the patient does not require oxygen supplementation or requires either > 50% FIO2 (as per conversion tables) or intubation.
2.6. Eligibility criteria: Inclusion criteria: (1) Male or non-pregnant female patients, age 18 years, that are confirmed COVID-19 positive by RT-PCR; and (2) Diagnosis of pneumonia requiring 21%<FIO2  50% to maintain saturation by pulse oximetry (SpO2) 90%; Exclusion criteria: (1) Respiratory decompensation requiring mechanical ventilation; (2) Hemodynamic instability requiring vasopressors; (3) Inability to maintain a sitting position during treatment; (4) Inability to effectively understand and communicate with the hyperbaric operator, or to give consent; (5) Contraindications to HBOT (e.g. pneumothorax); (6) Inability to spontaneously equalize ears and refusal of myringotomies; (7) Already enrolled in a conflicting research study.
2.7. Outcomes: The outcomes selected are reliable, patient-centered and meaningful to the healthcare system. Patients will be followed daily during hospitalization until Day 28 upon hospital discharge or death.


Primary Outcome: The 7-level COVID Ordinal Outcomes Scale assessed on Day 7 post- randomization, defined in Table 1. This outcome scale was used in other COVID studies28 and was also recommended by the WHO R&D Blueprint expert group.29 Ordinal scales have been used as end points in clinical trials for other viral pneumonia.29 32 We have chosen the ordinal-scaled outcome analyzed with a proportional odds logistic regression model to avoid the loss in statistical power and precision from dichotomization.
Secondary Outcomes: 1- Clinical outcomes: (1) Length of hospital stay; (2) Days with oxygen supplementation; (3) Daily oxygen flow values required to obtain saturation values 90%; (4) ICU admission, (5) ICU length of stay, (6) Days on invasive mechanical ventilation; (7) Major arterial and venous thrombotic events, e.g. stroke, pulmonary embolism, deep vein thrombosis; (8) Mortality; (9) Safety, defined as any adverse events related to HBOT.

2- Biological outcomes: plasmatic inflammation markers (CRP, D-Dimers, IL-6, IL-10, IL-12, IL-17, TNF, INF), measured at baseline and at D7 or hospital discharge, whichever comes first.

3- Cost of care, as measured by the gold standard Client Service Receipt Inventory (CSRI) at D28. The CSRI is adapted to participating countries to capture all social and healthcare care related expenses but also financial loss due to health.

2.8. Sample size: Unlike traditional frequentist approaches that propose a fixed sample size based on desired Type I and Type II error rates, a Bayesian approach does not require a fixed sample size in advance. Thus, we estimated the maximum anticipated sample size required using a frequentist approach. Our estimate of 234 patients randomized 1:1 treatment to control is adequate to achieve 80% power to detect an Odds Ratio (OR) of 2 using a proportional odds model and assuming the observed distribution of the ordinal scale at day 7 in the usual care arm of the Cao et al., New England Journal of Medicine COVID trial.28An OR of 2 corresponds to the following expected distribution at D7 (Table 1). We consider differences at level 1-3 to be clinically relevant (e.g., a difference from 7 to 3.6% mortality).

If information accruing during the trial differs from that expected, the sample size might change. Whereas unblinded sample size re-estimation in the frequentist setting requires multiplicity adjustment, the sequential Bayesian procedure will allow for early termination, as well as the possibility of study extension should results be promising but equivocal. Sample sizes needed to detect a range of ORs using a proportional odds model are as in Table 2.


2.9. Health services research issues: Healthcare resource use and economic outcomes will be measured prospectively using a modified version of Client Service Receipt Inventory.33 Health economic evaluation from a societal perspective will be performed alongside the trial to assess whether the benefits gained by HBOT outweigh its costs. The evaluation will be performed according to the best practices for economic evaluation in RCTs.34

2.10. Recruitment, compliance, and loss to follow up: The site Research Coordinator will identify patients admitted to the COVID-19 units and meeting eligibility criteria. The patient will be asked for their informed consent. With 5 recruiting centers across 3 countries, we are confident to reach our sample size. Since the beginning of May, both TOH (Ottawa) and Centenary Hospital (Toronto) have admitted over 30 non-intubated patients admitted for a COVID-19 pneumonia requiring oxygen supplementation. Evidence-based prediction for the coming months are between 1 and 4 new patients admitted daily at TOH.35 More COVID cases are expected in Edmonton in June-July. On May 11, The University Hospitals of Geneva had 185 non-ICU COVID patients admitted.36 The UK has the second largest number of COVID patients in the world. Our strong collaboration between hyperbaric medicine, internal medicine and intensive care (co-investigator team) and our multicentre design further limit the risks. We expect most patients will be compliant with HBOT based on our extensive experience with this treatment.

2.11. Study centers: We will recruit admitted patients from Ottawa (General Campus, TOH), Toronto (Centenary Hospital), Edmonton (Misericordia Community Hospital), Rugby/Coventry (UK) and Geneva, Switzerland (The Geneva University Hospitals).

2.12 Data Analysis

2.12.1. Primary outcome. Our primary analytical approach will use a Bayesian proportional odds (PO) ordinal logistic semiparametric model37 with prior distributions for the unknown parameters (Dirichletdistribution for the intercept and skeptical priors for treatment effect as recommended). The summary treatment effect is an Odd Ratio (OR) with OR > 1 being favorable to HBOT. The PO model will be adjusted for the following covariates: SOFA score; age; sex; the ordinal outcome scale measured at baseline; time from symptom onset in days; and factors known to be associated with poor outcomes (cardiovascular disease, immunocompromised, long-term care resident). Posterior probabilities will first begin to be computed after 20 patients have completed follow-up, and will be computed every other day thereafter. Example posterior probabilities (to be set by the Trial Steering Committee and DSMB) are presented in Table 3 and will be reported conditioned on all data accumulated to that point.

2.12.2 Secondary outcomes: Length of hospital stay will be analyzed using linear regression with log-transformation. Number of days with oxygen supplementation and days on invasive mechanical ventilation will be analyzed using Poisson or negative binomial regression. Incidence of major thrombotic events, mortality and adverse events related to HBOT will be analyzed using logistic regression or Fisher's exact test. Biological
outcomes will be analyzed using repeated measures generalized linear regression. Since evidence on COVID-19 evolves quickly, additional variables and interactions will be considered based on newer evidence published by the end of our trial.

2.13. Planned subgroup analyses: The primary and secondary outcomes will be analyzed for effect modification (treatment X subgroup interaction terms) by pre-specified factors that are known to have impact on COVID-19 prognosis and/or HBOT: biological sex and age.

2.14. Frequency of analyses, interim analyses: Safety and study-related adverse events will be reviewed regularly by an independent data safety monitoring board (DSMB). The DSMB will meet after 20 patients are completed and then every 2 weeks. The DSMB will recommend stopping the study if they find that one strategy is clearly harmful or beneficial. For example, the DSMB may recommend to stop with evidence for efficacy if P1>0.95; stop with evidence for moderate or greater efficacy if P2>0.8; stop with evidence for inefficacy if P4>0.8; stop with evidence for harm if P5>0.75.

3.0. Trial management: The trial's coordinating centre will be the Ottawa Methods Center (OMC) at the Ottawa Hospital Research Institute. The OMC has extensive experience leading and coordinating national and international multicenter trials. The Executive Committee (Principal Applicants and Patient Partner) will oversee trial management and will meet weekly. The DSMB will operate independently, but will act in an advisory role to the Executive Committee.

4.0. Roles of investigators: The HBOT for COVID-19 Trial investigators are an experienced, multidisciplinary team with substantial expertise in HBOT, clinical trials and knowledge translation. Nominated Principal Applicant (Dr. Boet) is a CIHR-funded investigator with experience conducting multicenter trials, has published widely and demonstrated his ability to lead complex projects. Co- Principal Applicant and lead Hyperbaric Medicine physician (Dr. Katznelson) has extensive experience in hyperbaric medicine and has multiple publications in this area. Applicant and lead Biostatistician (Dr. Taljaard) is a CIHR-Funded Senior Scientist and is an international expert in design, conduct and analysis of clinical trials, including large scale interventions. Co-Principal Applicant and lead Epidemiologist (Dr. Fergusson) is a CIHR-Funded Senior Scientist and international expert in clinical trials. Co-Principal Applicant and lead Internist (Dr. Castellucci) is a clinician scientist with expertise in clinical trials and treating COVID patients. Dr. J. MacDonald, Co- Principal Applicant Knowledge User is the president of the Canadian Undersea and hyperbaric Medicine Association (CUHMA). A strong team of patient partners, knowledge users, clinicians, biologists (Booth) and scientists, will make substantial contributions as a Steering Committee through monthly meetings. The Steering Committee includes patient partners (Proulx), experts in internal medicine (Castellucci), integrated knowledge translation (Boet, Clarke), health economics (Thavorn), sex/gender expert (Etherington), intensive care medicine (Djaiani), hyperbaric medicine (Pollock, Soibelman, Pignel, Blatteau, Wherret, Gagné, Moffet, Boet, Katznelson), administrative data (Sun), administrative leadership (Roth).