Effect of Remdesivir vs Standard Care on Clinical Status at 11 Days in Patients With Moderate COVID-19

Patients

Hospitalized patients with SARS-CoV-2 infection confirmed by polymerase chain reaction assay within 4 days of randomization and moderate COVID-19 pneumonia (defined as any radiographic evidence of pulmonary infiltrates and oxygen saturation >94% on room air) were enrolled. Patients with alanine aminotransferase or aspartate aminotransferase greater than 5 times the upper limit of normal or creatinine clearance of less than 50 mL/min were excluded. (Detailed eligibility criteria are provided in eAppendix 1 in Supplement 3.)

Study Design and Oversight

Patients were enrolled at 105 hospitals in the United States, Europe, and Asia between March 15, 2020, and April 18, 2020, and randomly assigned in a 1:1:1 ratio to receive up to a 5-day course of remdesivir, up to a 10-day course of remdesivir, or standard care. Randomization was not stratified. The randomization list was created and validated by the interactive web response system (IWRS) vendor. A dummy randomization list was provided in Microsoft Excel format to the biostatistician employed by the study sponsor for review. A separate list of sequential patient numbers within each treatment group was generated by the IWRS vendor. The randomization had a block size of 6. Based on the treatment from the randomization list, the IWRS provided the next sequential patient number to the site along with the treatment group assignment. The appropriate number of vials of open-label study drug were assigned to the patient. Sites did not have access to the randomization list and could not know the sequence of treatments. Treatment was open label because the sponsor had an insufficient number of placebo-containing vials to support this trial.

All patients randomized to a remdesivir group received 200 mg of remdesivir intravenously on day 1, followed by 100 mg of remdesivir once daily for the subsequent days, infused over 30 to 60 minutes. Remdesivir treatment was to be discontinued in any patient experiencing severe elevations in liver enzymes or decreases in estimated creatinine clearance to less than 30 mL/min (see eAppendix 1 in Supplement 3 for details). Patients who had sufficiently improved in the judgment of the investigator could be discharged from the hospital before finishing their assigned course of treatment.

The protocol was amended on March 15, 2020, on the basis of emerging understanding of the clinical presentation and assessment of COVID-19. The age limit for eligibility was lowered from 18 years to 12 years and the minimum temperature requirement was eliminated. The primary end point in the original protocol was the proportion of patients discharged by day 14 of the study. This was amended to assessment of clinical status on a 7-point ordinal scale by day 11 (described below).^11,14,16 In addition, a nonrandomized extension phase was added in which up to 1000 additional patients could be enrolled to receive remdesivir; the results of the extension phase will be the subject of a subsequent report. The statistical analysis plan was approved on May 25, 2020, prior to the database lock for the day 11 analysis. The statistical analysis plan was amended on June 26, 2020, prior to the database lock for the day 28 analysis. A detailed account of changes to the protocol is provided in eAppendix 1 in Supplement 3. All protocol amendments were authorized and approved by the sponsor, the institutional review board or independent ethics committee, and the pertinent regulatory authorities (eg, Food and Drug Administration for the United States; European Medicines Agency for Europe). The original protocol allowed use of other agents with presumptive activity against SARS-CoV-2 if such use was local standard care. Although this exception was disallowed in a subsequent amendment, some patients had already received other concurrent therapies.

Clinical and Laboratory Monitoring

Patient assessments included physical examination, respiratory status (respiratory rate, type of oxygen supplementation, blood oxygen saturation, and radiographic findings), adverse events, and concomitant medications. On study days 1, 3, 5, 8, 10, and 14, blood samples were obtained for measurement of blood cell counts, serum creatinine, glucose, total bilirubin, and liver transaminases. Self-reported fixed race and ethnicity categories were collected as part of the demographic information to assess possible differences in disease severity or response to treatment.

Site investigators assessed clinical status daily from day 1 through day 14 or hospital discharge on a 7-point ordinal scale^11,12 consisting of the following categories: 1, death; 2, hospitalized, requiring invasive mechanical ventilation or extracorporeal membrane oxygenation; 3, hospitalized, requiring noninvasive ventilation or use of high-flow oxygen devices; 4, hospitalized, requiring low-flow supplemental oxygen; 5, hospitalized, not requiring supplemental oxygen but requiring ongoing medical care (related or not to COVID-19); 6, hospitalized, not requiring supplemental oxygen or ongoing medical care; and 7, not hospitalized (eTable 1 in Supplement 3). If the clinical status of patients who remained hospitalized changed on a particular day, the worst score was documented. A final assessment was conducted on day 28 in person for hospitalized patients or by phone for those who had been discharged.

End Points

The primary efficacy end point was the distribution of clinical status assessed on the 7-point ordinal scale on study day 11. If remdesivir treatment improves outcomes, the distribution of scores among patients who received remdesivir should shift more toward the higher values of the scale than the distribution of scores among patients who received standard care.

The secondary end point was the proportion of patients with adverse events throughout the duration of the study. Prespecified exploratory end points were (1) time to recovery (improvement from a baseline score of 2-5 to a score of 6 or 7 or from a baseline score of 6 to a score of 7); (2) time to modified recovery (improvement from a baseline score of 2-4 to a score of 5-7, improvement from a baseline score of 5 to a score of 6-7, or improvement from a baseline score of 6 to a score of 7); (3) time to clinical improvement (≥2-point improvement from baseline on the 7-point ordinal scale); (4) time to 1-point or larger improvement; and (5) time to discontinuation of any oxygen support. The proportion of patients with these end points was also assessed on days 5, 7, and 11. Other exploratory end points were duration of hospitalization, duration of different modes of respiratory support, and all-cause mortality. Due to logistical issues at the time of study implementation, virological and pharmacokinetic measurements were limited, including SARS-CoV-2 polymerase chain reaction tests on day 5 and day 10, and are not presented.

Statistical Analysis

We calculated that 600 patients (200 in each group) would provide greater than 85% power to detect an odds ratio of 1.8 for each remdesivir group vs the standard care group using a 2-sided significance level of .05. The odds ratio of 1.8 was calculated based on proposed group sizes at the time of study conception and was not intended as a minimum clinically meaningful treatment effect, as no prior data were available on the distribution of clinical status categories over time in patients with moderate COVID-19. An odds ratio greater than 1 indicates changes in clinical status across all categories toward category 7 for the remdesivir groups vs the standard care group. All patients who were randomized and received at least 1 dose of remdesivir, or for the standard care group, had the day 1 visit, were assessed for efficacy and adverse events. For clinical status, the ordinal score was recorded as 1 on the day of death and all subsequent days; if a patient was discharged, the ordinal score was recorded as 7 on the day of discharge alive and all subsequent days unless the patient was rehospitalized for COVID-19–related reasons; otherwise, the most recent assessment was used for missing values. We used SAS version 9.4 (SAS Institute Inc) for all analyses.

For the primary efficacy end point, each remdesivir group was compared with the standard care group at a 2-sided α = .025 (Bonferroni). Proportional odds models were used with treatment as the independent variable; odds ratios and 95% CIs are presented. The assumption of proportional odds was tested using the score test, and supporting P values from the Wilcoxon rank sum test are provided if the proportional odds assumption was not met. Analyses including baseline clinical status as a covariate were also performed.

For the secondary end point of proportion of patients with adverse events throughout the duration of the study, comparisons between each remdesivir group and the standard care group were performed using a Fisher exact test; point estimates of the group differences and corresponding 95% CIs were calculated. For the prespecified exploratory end points, death was considered the competing risk in these time-to-event analyses. Patients without the event of interest were censored on the day of the last nonmissing ordinal scale assessment.

All-cause mortality was estimated using the Kaplan-Meier product limit method with all available data. Each remdesivir group was compared with the standard care group using the log-rank test, and hazard ratios and 95% CIs were provided. Participants who did not die were censored on the last study day.

Durations of oxygen therapy and hospitalization were summarized and compared between groups using the Wilcoxon rank sum test.

Post Hoc Analyses

We conducted sensitivity analyses of the primary end point (1) adjusting for day 1 clinical score; (2) adjusting for duration of symptoms; (3) using day 28 visit data to confirm day 11 clinical status and imputing patients with missing status as dead; and (4) using all randomized patients whether they received treatment or not (intention-to-treat population). We calculated the proportions of patients with a 1-point or greater improvement at day 11 within subgroups based on symptom duration, body mass index, race, baseline oxygen support, region, sex, and age. To understand if the open-label design and assigned duration of treatment had an effect on hospital discharge, we calculated rates of discharge in each group over time.

Because of the potential for type I error due to multiple comparisons, findings for analyses of end points other than the primary end point should be interpreted as exploratory.

Patient Disposition and Characteristics

Of 612 patients who consented and were assessed for eligibility, 596 underwent randomization and 584 began the study: 193 began a 10-day course of remdesivir, 191 patients began a 5-day course of remdesivir, and 200 continued standard care (Figure 1). Of the 16 patients who were not randomized, 13 did not meet eligibility criteria and 3 withdrew consent. Twelve randomized patients did not receive treatment: 8 withdrew consent, 3 had protocol violations, and 1 was withdrawn by investigator discretion.

Patients in the 3 groups were balanced in demographics and disease characteristics (Table 1). Overall, 56% of patients had cardiovascular disease, 42% had hypertension, 40% had diabetes, and 14% had asthma. Although all patients had an oxygen saturation above 94% while breathing room air at screening, 12% of patients in the 10-day remdesivir group, 16% in the 5-day remdesivir group, and 19% in the standard care group used supplemental oxygen on day 1 because of deteriorating clinical status or for breathing comfort. Patients in all groups had been hospitalized a median of 2 days (interquartile range [IQR], 1-3 days) before study day 1. The median duration of symptoms before day 1 in the standard care group was 9 days (IQR, 6-11 days) compared with 8 days (IQR, 5-11 days) for both groups receiving remdesivir. Patients randomized to the standard care group were more commonly prescribed other agents with putative activity against COVID-19 (Table 1).

Of 191 patients in the 5-day remdesivir group, 145 (76%) completed the assigned treatment duration (median, 5 doses; range, 1-5) (eTable 2 in Supplement 3). Reasons for discontinuing treatment included hospital discharge (35 patients [18%]), withdrawal of consent (5 [3%]), and adverse events (4 [2%]). Of the 193 patients in the 10-day remdesivir group, 73 (38%) completed the assigned treatment duration; the median number of doses for the group was 6 (range, 1-10). Reasons for discontinuing treatment included hospital discharge (98 patients [51%]), adverse events (8 [4%]), and withdrawal of consent (6 [3%]) (Figure 1). Five hundred thirty-three patients (91%) completed the study through day 28.

There were 37 patients (6.3%) who had not died, had not been discharged, and did not have a clinical status reported on day 11: 27 were transferred to another facility before day 11, 6 withdrew consent, 1 was withdrawn for a protocol violation, and 3 were discharged prior to day 11 but rehospitalized after day 11. The last available clinical status was used to impute clinical status on day 11 for these 37 patients.

Efficacy

Adverse Events (Secondary End Point)

Adverse events were experienced by 51% of patients in the 5-day remdesivir group, 59% in the 10-day remdesivir group, and 47% in the standard care group (Table 3). The difference in proportions between the 5-day remdesivir group and standard care was not statistically significant (4.8%; 95% CI, –5.2% to 14.7%; P = .36), but the difference between the 10-day remdesivir group and standard care was significant (12.0%; 95% CI, 1.6%-21.8%; P = .02). Adverse events that were more common in the remdesivir groups than in the standard care group include nausea, hypokalemia, and headache (Table 3).

Serious adverse events were less common in the remdesivir groups (5% in both) than in the standard care group (9%), differences of −4.3% (95% CI, −9.7% to 0.9%; P = .11) for the 5-day remdesivir group vs standard care and −3.8% (95% CI, −9.3% to 1.4%; P = .17) for the 10-day remdesivir group vs standard care. All 9 deaths through day 28 (2 [1%] in the 5-day remdesivir group, 3 [2%] in the 10-day remdesivir group, and 4 [2%] in the standard care group) occurred in patients aged 64 years or older, and none was attributed to remdesivir treatment.

Limitations

This study has several limitations. First, the original protocol was written when COVID-19 cases were largely confined to Asia and the clinical understanding of disease was limited to case series.^19,20 This led to a change in the primary end point on the first day of study enrollment as it became clear that hospital discharge rates varied greatly across regions and the ordinal scale had become standard for interventional COVID-19 studies.^11,21 Second, the study used an open-label design, which potentially led to biases in patient care and reporting of data. Third, because of the urgent circumstances in which the study was conducted, virologic outcomes such as effect of remdesivir on SARS-CoV-2 viral load were not assessed. Fourth, other laboratory parameters that may have aided in identifying additional predictors of outcomes were not routinely collected. Fifth, the ordinal scale used to evaluate outcomes was not ideal for detecting differences in patients with moderate COVID-19, especially for a clinical situation in which discharge decisions may be driven by factors other than clinical improvement.