This phase Ib/IIa study is an investigator initiated, prospective randomised, parallel-group, multi-centre open label, proof of principle, superiority study. The study took place across three intensive care units (ICUs) in Ireland. As a phase Ib/IIa study, the co-primary endpoints were safety outcomes and efficacy as evidenced by changes in the d-dimer profile in recruited patients.

The study was registered with the EU clinical trials database (2020-003349-12) and clinicaltrials.gov (NCT04511923), registered 11/08/2020. Ethics approval was obtained from the National Research Ethics Committee (NREC) in Ireland, approval number 20-NREC-COV-104. Regulatory approval for the study was obtained from the Health Products Regulatory Authority, approval number CT0900/650/001 Heparin Sodium. The protocol and statistical analysis plan for this study has been peer reviewed and published previously [10].

Inclusion criteria for this study were as follows: (1) confirmed or suspected COVID-19 (if ‘suspected’, results must be pending or testing intended); (2) ability to obtain informed consent/assent to participate in study; (3) age 18 years or older; (4) requiring high-flow nasal oxygen (> 30L/min) or positive pressure ventilator support or invasive mechanical ventilation in the ICU for a time period of no greater than 48 h; (5) d-dimers > 200 ng/ml; (6) acute opacities (not effusions, lobar/lung collapse or nodules) on chest imaging affecting at least one lung quadrant; (7) currently in a higher level of care area designated for inpatient care of patients where advanced respiratory support therapies can be provided.

Exclusion criteria included the following: Enrolled in another clinical trial that was unapproved for co-enrolment, heparin allergy or heparin-induced thrombocytopenia, APTT > 100 s, pulmonary bleeding, platelet count < 50 × 109 per L, uncontrolled bleeding, pregnancy, technical/equipment constraints, clinician objection, the use or anticipated use of nebulised tobramycin during this clinical episode, any other specific contraindication to anticoagulation, receiving any direct/novel oral anticoagulant. The full exclusion criteria are described in detail in the methods paper [10].

Patients meeting the eligibility criteria were approached for consent to participate, or in the case of patients incapable of providing consent, deferred consent and relative assent to participate were sought. The process was approved by the Health Research Consent Declaration Committee in Ireland.

Allocation was carried out after consent or assent were confirmed, via a central, secure web randomisation process embedded in the eCRF. The eCRF was hosted by the contract research organisation (CRO, Afortiori Development, Galway, Ireland). Allocation was in a one-to-one ratio, with variable block size randomisation. The Programme R was used to generate the randomisation sequence. Blocks of variable size and a random seed were used to ensure allocation concealment. Block size was not revealed to site investigators by the CRO. Site-level randomisation was used. Participants were assigned to receive standard ICU care plus nebulised heparin or standard care alone.

In the treatment group, nebulised unfractionated heparin 25,000 units (5 ml heparin sodium 5000 IU/ml (Pinewood laboratories, Clonmel, Ireland) was administered via the Aerogen Solo® nebuliser every 6 h from enrolment to day 10, or until discontinuation of advanced respiratory support (if sooner), with the dosage and schedule based on previous work [1, 11, 12]. The control standard of care (SOC) group received standard care.

Outcomes are based on data collected up to day 10 for laboratory and some clinical results, with other clinical endpoints and safety and adverse event data collected up to day 60. All outcomes and timepoints are detailed in the methods paper [10].

Primary outcomes. The co-primary outcomes were a between-group difference in d-dimer concentrations over time, and a between-group difference in the occurrence of serious adverse events (SAEs) as listed in the safety analysis.

Secondary outcomes. Secondary efficacy outcomes include analysis of between-group differences in: P/F ratio and pulmonary compliance (in invasively ventilated patients). The ROX index, defined as the ratio of oxygen saturation as measured by pulse oximetry/FIO2 to respiratory rate was assessed in patients receiving high-flow nasal oxygen (HFNO) therapy. The oxygenation index (OI = mean airway pressure MAP × FiO2 × 100 ÷ PaO2) was assessed in patients receiving positive pressure ventilation. The time to separation from invasive ventilation to day 28, number tracheotomised to day 28 (N), time to ICU discharge, and survival (day 28, day 60 and hospital discharge) were also assessed. Inflammatory and coagulation indices included interleukins (IL-1β, IL-6, IL-8, IL-10), soluble tumour necrosis factor receptor 1, C-reactive protein, procalcitonin, ferritin, fibrinogen and lactate dehydrogenase), ratio of IL-1β/IL-10 and IL-6/IL-10.

Safety outcomes. These include analysis of between-group differences in: patients transfused packed red blood cells (PRBCs); volume of PRBCs transfused; major bleeding, (defined as bleeding that results in death and/or bleeding that is symptomatic and occurs in a critical area or organ); bleeding that results in a fall in haemoglobin of 20 g/l or more, or results in transfusion of > 2 units of PRBCs; number of patients who record clinically relevant non-major bleeding as defined by the International Society on Thrombosis and Haemostasis [13], number developing heparin-induced thrombocytopenia (HIT), number of other adverse events and reactions, all to day 10. AEs were classified by organ system. Safety data on air quality samples from an infection control perspective will be published separately.

Sample size. As a phase Ib/IIa study, and as published data relating to COVID 19 on which to power this study were lacking at that time, we based our sample size calculation on in-house data for d-dimer levels in COVID-19 patients who required ICU care (12 patients) and those who did not (26 patients). Based on an estimation that nebulised heparin may reduce the d-dimer levels in ICU patients (mean = 944.8 ng/ml [SD = 485.3]) to those experienced in ward patients (mean = 436.5 ng/ml [SD = 604.0]), with an alpha level of 0.05 and a power of 90% to detect a type II error, 19 patients per group were required. Increasing the number by 1 per group to allow for potential dropout gave a sample size of 20 patients per arm of the study, with a total of 40 patients to be enrolled. The safety analysis was not used to power the study, but the safety outcomes were analysed as a co-primary outcome.

Demographic data were summarised appropriately by treatment arm. The primary outcomes analysis was carried out on an intention to treat basis. Safety analysis was carried out also, including all recruited patients.

Effectiveness analysis for the primary outcome was carried out using analysis of covariance for d-dimer AUC, adjusting for baseline AUC, and Fisher’s exact test for safety outcomes. Continuous outcomes were statistically compared between treatment groups using t-tests or Mann–Whitney U tests (for non-normal variables). Linear mixed effects models, were used for the analysis of longitudinal data, representing the effect of heparin by differential slopes of response trajectories over time between the heparin and control arms. Responses were log-transformed, prior to fitting these models, when distributions were notably skewed. A Cox proportional hazards model was used to model time to event responses. Interval estimates (95% confidence intervals) for the treatment effect were reported in addition to the point estimates, for all response variables [10].