Figure 1 displays the value tree of the key benefits and risks related to remdesivir treatment in COVID-19. The benefits included in the value tree include key endpoints from clinical trial protocols for studies assessing the efficacy of remdesivir in severe COVID-19 disease. As remdesivir is currently not approved for use in any condition, it is acknowledged that its safety profile has not been completely characterised. For the purposes of identifying potential risks associated with the use of remdesivir in COVID-19 disease for the value tree, safety data has been identified from currently available sources that met the inclusion/exclusion criteria. These include studies reporting its use in the treatment of Ebola virus disease , case series documenting the use of remdesivir in COVID-19 disease , and safety data included in the study by Wang et al. . Both the efficacy and safety outcomes have been presented in ranked numerical order according to perceived clinical significance. The key characteristics of clinical trials for remdesivir in any population have been presented in Table 1. As per the BRAT method, only studies that met the inclusion/exclusion criteria for the current disease context (severe COVID-19) were included in the data extraction phase.
The benefits listed in the value tree represent key clinical endpoints included in clinical trial protocols. These have been ranked in order of perceived clinical importance, from the clinical endpoint of mortality risk through to the surrogate endpoint of viral clearance. Many studies have utilised ordinal scales, which include a spectrum of the clinical status of the patient. The primary endpoint used in the recently published study by Wang et al.  was time to clinical improvement up to Day 28; this was defined as the time (in days) from randomisation to the point of a decline of two levels on a six-point ordinal scale of clinical status (from 1 = discharged to 6 = death) or discharged alive from hospital, whichever came first. The six-point scale was as follows: death = 6; hospital admission for extracorporeal membrane oxygenation or mechanical ventilation = 5; hospital admission for non-invasive ventilation or high-flow oxygen therapy = 4; hospital admission for oxygen therapy (but not requiring high-flow or non-invasive ventilation) = 3; hospital admission but not requiring oxygen therapy = 2; and discharged or having reached discharge criteria = 1.
The primary endpoint in the Multicentre, Adaptive, Randomized Blinded Controlled Trial of the Safety and Efficacy of Investigational Therapeutics for the Treatment of COVID-19 in Hospitalized Adults, known as the Adaptive COVID-19 treatment trial sponsored by the National Institute of Allergy and Infectious Diseases (NIAID) , was time to recovery (time frame: Day 1 through Day 29). Day of recovery was defined as the first day on which the subject satisfied one of the following three categories from the ordinal scale: (1) hospitalized, not requiring supplemental oxygen—no longer requires ongoing medical care; (2) not hospitalized, limitation on activities and/or requiring home oxygen; (3) not hospitalized, no limitations on activities. Other key clinical outcomes included duration of ventilation, duration of oxygen support, and duration of hospital stay, in addition to time from randomization to discharge or death. Surrogate endpoints assessing viral load clearance are likely to be less robust forms of endpoint data, and accordingly have been ranked lower in terms of clinical importance.
As remdesivir is an unapproved medicine in most countries, its safety profile has not yet been fully characterised. Known key risks at the current time have been included in the value tree, identified from a variety of sources, including recently published clinical trial data in the context of COVID-19 . These have also been ranked according to perceived seriousness.
Cardiovascular outcomes including hypotension and arrhythmias have been documented following the use of remdesivir [8, 26, 29]; however, the risk of cardiovascular outcomes with remdesivir remains largely unknown . Hypotension was reported in one patient in the phase III study investigating remdesivir in the context of Ebola virus disease; this patient subsequently suffered a cardiac arrest, although the manuscript states that this death could not readily be distinguishable from underlying fulminant Ebola virus disease itself . Cardiac arrest was also reported in a further patient in the remdesivir group reported in the COVID-19 study by Wang et al. . Multiple organ dysfunction, septic shock, acute kidney injury and hypotension have also been reported as adverse events amongst patients provided with remdesivir either on a compassionate-use basis, or in a clinical trial [26, 27]. Respiratory failure or acute respiratory distress syndrome has been cited as an adverse event in patients taking remdesivir [27, 31], and as such has been included here, although it is acknowledged that this may be related to underlying disease (COVID-19), rather than to remdesivir. Elevations in liver transaminases and gastrointestinal events, including diarrhoea, have also been reported with the use of remdesivir [8, 13, 26, 32], in addition to reports of haemorrhage of the lower gastrointestinal (GI) tract [27, 33].
Where available, clinical trial data relating to any efficacy or safety outcomes included in the value tree for both remdesivir and a comparator group in the context of severe COVID-19 were extracted and included in Table 2. Based on this data, summary metrics including risks per 1000 patient years and risk differences are presented in Table 3. We identified 68 papers through literature searching from PubMed and 384 papers from Google Scholar. We also identified results from one clinical trial on the NIAID website. After initial review, one paper  and results from one clinical trial on the NIAID website  were included in the final benefit–risk assessment.
In the Wang et al. trial, the benefit of time to clinical improvement was not statistically significant (21 vs 23 days, HR 1.23, 95% CI 0.87–1.75). In the NIAID trial, a statistically significant shorter time to recovery in patients treated with remdesivir was observed (11 vs 15 days; p < 0.001), with non-significant reduced mortality risk (8% vs 12%; p = 0.059). Other non-significant benefit data were identified from the Wang et al. trial, including invasive ventilation and oxygen use at Day 28 (1% vs 4% and 12% vs 17%, respectively).
Risk data were only available from the Wang et al. trial, which reported fewer serious adverse events in patients taking remdesivir (18%) compared with the placebo group (26%); however, more patients in the remdesivir group discontinued treatment as a result of an adverse event compared with those patients receiving placebo (12% vs 5%).
A serious adverse event is any event that results in death, is life-threatening, requires hospitalisation or prolongation of existing hospitalisation, results in persistent or significant disability or incapacity, or is a birth defect.