Randomized Controlled Trial
Study population and setting
The ACTT-2 study involved 1,033 adults hospitalized with COVID-19 admitted to 67 hospitals in 8 countries between May 8 and July 1, 2020. At each hospital, patients were randomized in equal proportions to either combination therapy (remdesivir plus baricitinib) or control (remdesivir plus placebo). The dose of baricitinib was adjusted for glomerular filtration rate (GFR). Inclusion required a positive SARS-CoV-2 RT-PCR test, and evidence of lower respiratory tract infection based on radiographic infiltrates, SpO2 ≤94% on room air, or requirement of supplementation oxygen. Exclusion criteria were hepatic injury (ALT or AST levels at >5 times the upper limit of normal), renal injury (estimated GFR>30 mL/min/mm2), need for hemodialysis or hemofiltration, contra-indication to study drug, pregnancy or breast-feeding, or anticipated transfer elsewhere within 72 hours. The primary outcome was time to recovery during the 28 days after enrollment using the World Health Organization Ordinal Scale for Clinical Improvement, with recovery defined as the first day on which an individual achieved categories 1, 2, or 3 on the Ordinal Scale. A major secondary outcome was clinical status at 15 days post-enrollment.
Summary of Main Findings
In intention-to-treat analysis, addition of baricitinib to remdesivir therapy reduced median time to recovery from 8 days to 7 days (rate ratio 1.16, 95% CI: 1.01 to 1.32, p 0.03). Among participants who were category 6 on the Ordinal Scale at baseline (receiving non-invasive ventilation or high-flow oxygen), the median time to recovery was 10 days for baricitinib compared with 18 days for control (rate ratio 1.15, 95% CI: 1.10 to 2.08), but this difference was not significant in patients with either less or more severe disease. In secondary analysis, improvement in clinical status at day 15 was greater for the baricitinib arm (odds ratio 1.3, 95% CI: 1.0 to 1.6), and the highest benefit was observed for patients who were category 6 on the Ordinal Scale at baseline (odds ratio 2.2, 95% CI: 1.4 to 3.6). The incidence of progression to death or ventilation was lower in the baricitinib arm compared to control (22.5% vs. 28.4%, 95% CI: 0.60 to 0.98). The baricitinib arm also had fewer serious adverse effects (16.0% vs. 21.0%; p 0.03) and new infections (5.9% vs. 11.2%; p 0.003) compared to control. There was no significant difference in 28-day mortality between patients receiving baricitinib or placebo (5.1% vs. 7.8%; hazard ratio 0.65, 95% CI 0.39 to 1.09).
This study assigned patients to combination therapy vs. control, with randomization stratified according to disease severity and trial site. The design was double-blind and placebo-controlled, meaning the participants and study teams were blinded to treatment, to minimize reporting bias. A diverse patient population was recruited from hospital sites across multiple countries, enhancing the generalizability of findings. In addition, a large proportion of Hispanic or Latino patients (51.4%) was enrolled in this study, an important strength given how disproportionately COVID-19 has affected these communities. Analyses were also stratified by sex, race, ethnic group, duration of symptoms before hospitalization, site location, and pre-existing conditions, allowing for detailed analysis of patient subgroups most amenable to baricitinib addition. Importantly, most sub-groups continued to show benefit of the addition of baricitinib, although not with statistical significance.
While recruitment from multiple hospitals may enhance generalizability of results, this led to heterogeneity in standard of care protocol. For hospitals with a written policy for COVID-19 treatments, patients could receive these additional treatments in addition to the experimental intervention. The prevalence of these additional interventions in each group was not specified in the results, and consequently there were no adjustments for elements that constituted a given standard of care. Additionally, there was a high loss of patients to follow-up in both baricitinib and placebo arms (40 vs. 41), although this was balanced between the two arms of the study. It is worth noting that there was an imbalance of participants in the placebo arm with no other comorbidities compared to baricitinib arm (18.3% vs. 12.9%). Median time of symptom onset to enrollment was relatively long at 8 days for each arm, precluding analysis of the benefits of short-term administration of baricitinib. Since most patients with early COVID-19 are not admitted to the hospital, there is likely to be significant bias in enrolling patients at later stages of disease. While there were fewer short-term adverse effects in the baricitinib arm, follow-up for patients was ceased at 29 days, precluding analysis of long-term effects of drug treatment or all-cause mortality. Importantly, the combined outcome of mortality and use of ventilation was not pre-specified, so statistically significant results may have arisen by chance. Multiple secondary outcomes, including clinical status, use of oxygen, use of ventilation, mortality, adverse effects, and duration of hospitalization, were measured, and the analyses were not corrected for multiple comparisons. In addition, since dexamethasone has recently been shown to improve mortality of persons with COVID-19, it would have been important to adjust the risk models for co-administration of for all corticosteroids in general and for dexamethasone specifically.
This study provides valuable descriptive data on occupational groups most impacted by COVID-19 over the course of the pandemic.