The authors respond to “Concerns about subgroup analyses and reason for stopping remdesivir trial”

We thank Dr. Vijayaraghavan for the letter¹ in response to our article.² The Solidarity trial (NCT04647669) stopped enrolment of patients on Jan. 29, 2021, because of a lack of available product with which to continue randomization at a rapid pace, not an interim analysis or data safety monitoring board (DSMB) recommendation. The Canadian Treatments for COVID-19 (CATCO) trial (NCT04330690) is an independent national trial with an independent steering committee and DSMB; drug was supplied through available hospital supply. The CATCO trial continued to enrol patients until Apr. 1, 2021, because without new data on effectiveness of the study medication from either Solidarity or another trial, equipoise remained, there were substantial numbers of patients with COVID-19 in Canada still eligible for the trial, and there was no information upon which the steering committee or DSMB could yet recommend stopping enrolment to the remdesivir component. Enrolment continued in CATCO until the Solidarity trial had a data set that was approaching readiness for analysis, and a decrease in hospital admissions due to COVID-19 in Canada.

We reported no statistically significant difference in the outcome of in-hospital mortality between study groups (19% for those treated with remdesivir, 23% for those in the control arm, relative risk [RR] 0.83, 95% CI 0.67–1.03).² It is possible that this effect is limited by lack of power and will be better clarified in the full Solidarity trial when it is published. There was a difference in the secondary outcomes of need for new mechanical ventilation, oxygen-free days at day 28 and ventilator-free days at day 28. At the sample size achieved, it is unlikely that these were due to chance and these are consistent with the final report of the ACTT-1 (the Adaptive COVID-19 Treatment Trial) study.³ We evaluated the influence of patient characteristics, duration of illness and baseline severity of illness on the primary outcome of in-hospital mortality in Figure 3 of the primary publication.² Although the difference in mortality was not statistically significant, there was a consistent nonstatistically significant signal of lower odds of dying among those treated with remdesivir in most sub-groups, including time-from-symptom onset at enrolment. As the largest trial to report the impact of duration of symptoms on outcomes (including the larger Solidarity trial, which did not collect this variable), we believe the lack of difference across duration is an important finding. Future post hoc analyses will look at symptom duration as a continuous variable to determine whether there is any association with very early or very late treatment with remdesivir efficacy.