
Table of Contents
Introduction
- Huang C
- Wang Y
- Li X
- et al.
,
- Ruan Q
- Yang K
- Wang W
- Jiang L
- Song J
,
,
Despite treatment advances with remdesivir, dexamethasone, and tocilizumab, reducing mortality among hospitalised patients remains a crucial unmet need.
- Beigel JH
- Tomashek KM
- Dodd LE
- et al.
,
- Kalil AC
- Patterson TF
- Mehta AK
- et al.
,
- Horby P
- Lim WS
- Emberson JR
- et al.
,
Tocilizumab in patients admitted to hospital with COVID-19 (RECOVERY): a randomised, controlled, open-label, platform trial.
- Fridman JS
- Scherle PA
- Collins R
- et al.
,
- Shi JG
- Chen X
- Lee F
- et al.
,
- McInnes IB
- Byers NL
- Higgs RE
- et al.
with a known anti-inflammatory profile in patients with autoimmune diseases.
- Sanchez GAM
- Reinhardt A
- Ramsey S
- et al.
,
- Tanaka Y
- Emoto K
- Cai Z
- et al.
,
- Dörner T
- Tanaka Y
- Petri MA
- et al.
In February, 2020, baricitinib was identified by an artificial intelligence platform as a potential intervention for the treatment of COVID-19 because of its known anticytokine properties and potential for targeting host proteins for its antiviral mechanism.
- Richardson P
- Griffin I
- Tucker C
- et al.
,
- Stebbing J
- Phelan A
- Griffin I
- et al.
The biochemical inhibitory effects of baricitinib on human Numb-associated kinases (AAK1, BIKE, and GAK) responsible for SARS-CoV-2 viral propagation were subsequently confirmed.
- Stebbing J
- Krishnan V
- de Bono S
- et al.
Baricitinib was also shown to reduce multiple cytokines and biomarkers implicated in COVID-19 pathophysiology.
- Bronte V
- Ugel S
- Tinazzi E
- et al.
,
- Petrone L
- Petruccioli E
- Alonzi T
- et al.
,
- Sims JT
- Krishnan V
- Chang CY
- et al.
Following the publication of these findings, several observational studies including small cohorts of hospitalised patients with COVID-19 (including older adults) were done and provided the first evidence of clinical improvement associated with baricitinib treatment.
- Titanji BK
- Farley MM
- Mehta A
- et al.
,
- Cantini F
- Niccoli L
- Matarrese D
- Nicastri E
- Stobbione P
- Goletti D
,
- Cantini F
- Niccoli L
- Nannini C
- et al.
,
- Stebbing J
- Sánchez Nievas G
- Falcone M
- et al.
Evidence before this study
We searched PubMed using the terms “COVID-19”, “SARS-CoV-2”, “treatment”, “baricitinib”, and “JAK inhibitor” for articles in English, published until April 31, 2020, regardless of article type. We considered previous and current clinical trials of investigational medications in COVID-19, as well as previous clinical trials of the Janus kinase (JAK)1 and JAK2 inhibitor baricitinib conducted before this study. At the time the COV-BARRIER study was designed, there were no approved therapies for the treatment of COVID-19. Management of COVID-19 was supportive, and few phase 3 randomised placebo-controlled studies had been completed. Some phase 2 and 3 data on the antimalarial hydroxychloroquine and protease inhibitor lopinavir–ritonavir were available, and trials investigating the use of the antiviral remdesivir were ongoing. Baricitinib was identified as a potential intervention for COVID-19 due to its mechanism of action as a JAK1 and JAK2 inhibitor, its known anticytokine properties, and a potential antiviral mechanism via targeting host proteins. Additionally, early case series evaluating the efficacy and safety of baricitinib in populations of hospitalised patients supported further evaluation of baricitinib as a potential treatment option for hospitalised patients with COVID-19. While the COV-BARRIER study was enrolling, ACTT-2, a phase 3 study evaluating baricitinib plus remdesivir, was completed and showed that baricitinib in combination with remdesivir improved time to recovery and other outcomes.
Added value of this study
This was the first phase 3 study to evaluate baricitinib in addition to the current standard of care, and included patients receiving antivirals, anticoagulants, and corticosteroids. After the earliest publication of the RECOVERY study in June, 2020, the treatment of hospitalised patients with COVID-19 changed with the adoption of dexamethasone as the standard of care. As a result of its design, COV-BARRIER became the first trial to evaluate the benefit and risk of baricitinib when added to the most current standard of care (dexamethasone) in these patients. This was a randomised, double-blind, placebo-controlled trial conducted globally in regions with high COVID-19 hospitalisation rates. The reduction in the composite primary endpoint of progression to non-invasive ventilation, high-flow oxygen, invasive mechanical ventilation, or death for baricitinib plus standard of care (including dexamethasone) compared with placebo plus standard of care was not statistically significant. However, analysis of a prespecified key secondary endpoint showed that treatment with baricitinib reduced 28-day all-cause mortality by 38·2% compared with placebo (HR 0·57 [95% CI 0·41–0·78], nominal p=0·0018), with one additional death prevented per 20 baricitinib-treated participants. The reduction of all-cause mortality with baricitinib was maintained up to day 60 in an exploratory analysis. The frequency of serious adverse events, serious infections, and venous thromboembolic events was similar between the baricitinib and placebo groups.
Implications of all the available evidence
In this phase 3 trial, baricitinib administered in addition to standard of care (which predominantly included dexamethasone) did not reduce the incidence of a composite endpoint of disease progression, but showed a strong effect on reduction of mortality by 28 days, an effect which was maintained up to 60 days. In the ACTT-2 study, baricitinib further reduced time to recovery above the background use of remdesivir. Taken together, these findings suggest that baricitinib has synergistic effects with other standard-of-care treatment modalities, including remdesivir and dexamethasone. Based on all available evidence, baricitinib is a potentially effective oral treatment option to decrease mortality in hospitalised patients with COVID-19.
- Kalil AC
- Patterson TF
- Mehta AK
- et al.
The US Food and Drug Administration issued an emergency use authorisation for baricitinib in hospitalised patients with COVID-19 who required oxygen supplementation,
Letter of authorization: EUA for baricitinib (Olumiant) for treatment of coronavirus disease 2019 (COVID-19).
addressing an unmet need in the treatment of COVID-19.
Globally, however, evaluations of new treatment options to reduce mortality in hospitalised patients with COVID-19 are still urgently needed to reduce the high frequency of deaths that persists despite improvements in standards of care.
The COV-BARRIER study was designed to evaluate the efficacy and safety of baricitinib in combination with standard of care, including dexamethasone, for the treatment of hospitalised adults with COVID-19. To the best of our knowledge, this study is the first double-blind, placebo-controlled trial to evaluate mortality by day 60.
Methods
Study design and participants
Following the disclosure of results from the ACTT-2 study showing that disease progression was unlikely in participants without baseline oxygen support, the COV-BARRIER protocol was amended on Oct 20, 2020, to limit enrolment to participants who required baseline oxygen support (NIAID-OS score 5 or 6).
Randomisation and masking
Participants were enrolled by study investigators (or a designee). Randomisation was facilitated by a computer-generated random sequence using an interactive web-response system, and was permitted by a study investigator or designee to allocate participants 1:1 to the baricitinib group or the placebo group. Participants were stratified according to the following baseline factors: disease severity (NIAID-OS 4, 5, or 6), age (<65 or ≥65 years), region (Europe, USA, or the rest of the world), and use of corticosteroids for primary study condition (yes or no). Participants, study staff, and investigators were masked to the study assignment. An independent, external data monitoring committee oversaw the study and evaluated unblinded interim data for efficacy, futility, and safety. An independent, blinded, clinical event committee adjudicated potential venous thromboembolic events and deaths.
Procedures
- Horby P
- Lim WS
- Emberson JR
- et al.
but higher corticosteroid doses (>20 mg per day [or prednisone equivalent] administered for >14 consecutive days in the month before study entry) were not permitted, unless indicated per standard of care for a concurrent condition, such as asthma, chronic obstructive pulmonary disease, or adrenal insufficiency. Prophylaxis for venous thromboembolic events per local practice was required for all participants unless there was a major contraindication, such as an active bleeding event or history of heparin-induced thrombosis.
Interventions were packaged in identical bottles containing tablets of either 2 mg baricitinib or matching placebo. The baricitinib intervention consisted of baricitinib at a dose of 4 mg/day; however, 2 mg/day was given if the patient had a baseline eGFR of 30 to less than 60 mL/min/1·73 m2. Baricitinib or placebo tablets were administered orally (or crushed for nasogastric tube delivery) and given daily for up to 14 days or until discharge from hospital, whichever occurred first. The date and time of each dose administered was recorded in the source documents and the case report form, and the site investigator (or designee) was responsible for assessing study drug compliance.
For efficacy and health outcomes, baseline measurements were defined as the last non-missing assessment recorded on or before the first administration of study drug at study day 1 (randomisation).
Participants followed the study visit schedule per the protocol, which included a study visit at day 28 for assessment of the primary endpoint. Efficacy and safety were evaluated for all participants up to day 28, and and all-cause mortality was also evaluated in participants with non-missing baseline and at least one post-baseline observation up to day 60. Participants had a follow-up visit approximately 28 days after receiving their last dose of study drug.
Outcomes
The composite primary endpoint was the proportion of participants who progressed to high-flow oxygen or non-invasive ventilation (NIAID-OS score 6), invasive mechanical ventilation or extracorporeal membrane oxygenation (NIAID-OS score 7), or death (NIAID-OS score 8) by day 28, in the baricitinib group compared with the placebo group. All-cause mortality by day 28 was a prespecified key secondary endpoint, and all-cause mortality by day 60 was a prespecified exploratory endpoint. The primary, intention-to-treat analysis was done in two populations: population 1 (comprising all randomised participants) and population 2 (the subpopulation of participants who, at baseline, required oxygen supplementation and were not receiving systemic corticosteroids for COVID-19).
Statistical analysis
of the primary endpoint, as assessed during an interim analysis evaluated by an external data monitoring committee (completed in January, 2021, with no changes recommended). Power calculations assumed that 75% of the total α was allocated to population 1, and that 60% of the participants were taking systemic corticosteroids at baseline. Two scenarios were considered. In the first, both population 1 and population 2 had a true treatment effect size of 7·5% (power 81%). In the second, population 1 had a true effect size of 4% and population 2 had an assumed effect size of 7·5% (power 54%). In the final (pre-unmasking) version of the statistical analysis plan, the total α was amended to be allocated 99% to population 1, recognising that population 2 was much smaller than previously anticipated and unlikely to succeed.
Efficacy data were analysed in the intention-to-treat population, defined as all randomly allocated participants. Logistic regression was used for dichotomous endpoints, proportional odds models were used for ordinal endpoints, ANOVA was used for continuous endpoints, and mixed-effects models of repeated measures were used to assess continuous endpoints over time. Log-rank tests and hazard ratios (HRs) from Cox proportional hazard models were used for time-to-event analyses. These statistical models were adjusted for treatment and baseline stratification factors. Prespecified subgroup analyses for the primary and selected key secondary endpoints evaluated treatment effect across the following subgroups: baseline severity (NIAID-OS score 4, 5, or 6), baseline systemic corticosteroid use (yes or no), baseline remdesivir use (yes or no), geographical region (Europe, USA, or the rest of the world), sex, disease duration at baseline (<7 days or ≥7 days), and age at baseline (<65 years or ≥65 years).
Role of the funding source
COV-BARRIER was designed jointly by consultant experts and representatives of the sponsor, Eli Lilly and Company. Data were collected by investigators and analysed by the sponsor. All authors participated in the interpretation of the data analysis, draft, and final manuscript review, and provided critical comment, including the decision to submit the manuscript for publication with medical writing support provided by the sponsor. The authors had full access to the data and authors from the sponsor verified the veracity, accuracy, and completeness of the data and analyses as well as the fidelity of this report to the protocol.
Results
Figure 1Trial profile
*159 deaths were reported by day 28; an additional three deaths occurred after the treatment period disposition but within 28 days.
Table 1Baseline demographics and clinical characteristics
Data are mean (SD) or n/N (%). NIAID-OS=National Institute of Allergy and Infectious Disease Ordinal Scale.
Table 2Primary and key secondary outcomes in the intention-to-treat population
Group data are %, n/N (%), median (95% CI), or least squares mean (SE). Population 1 includes all randomised participants. Population 2 includes participants who, at baseline, required oxygen supplementation and were not receiving dexamethasone or other systemic corticosteroids for the primary study condition. Data were assessed from days 1 to 28, unless otherwise indicated. Dichotomous endpoints were analysed with a logistic regression model. Ordinal efficacy endpoints were analysed with a proportional odds model. Continuous endpoints were analysed by ANOVA. All of these analyses had baseline randomisation factors and treatment group in the model, except in cases where the factor was redundant in the model (eg, for population 2, baseline corticosteroid use [yes or no] could not be included in the model because no participants in this population were on corticosteroids at baseline). For time-to-event endpoints, p values were calculated using an unstratified log-rank test. The HR (and corresponding 95% CI) was calculated using a Cox proportional hazards model. ECMO=extracorporeal membrane oxygenation. OR=odds ratio. HR=hazard ratio. NIAID-OS=National Institute of Allergy and Infectious Disease Ordinal Scale. RR=rate ratio. LSMD=least squares mean difference.
In population 2, 28-day all-cause mortality was 5% (five of 96 participants) in the baricitinib group and 15% (16 of 109) in the placebo group, equating to a 65% relative reduction (HR 0·31 [95% CI 0·11–0·88], nominal p=0·030; figures 2B, 3).

Figure 2Kaplan-Meier estimates of 28-day and 60-day all-cause mortality, and distribution of participants with each NIAID-OS score over time
(A–F) 28-day all-cause mortality in population 1, the overall population (A); population 2, comprising participants who, at baseline, required oxygen supplementation and were not receiving dexamethasone or other systemic corticosteroids for the primary study condition (B); populations with baseline NIAID-OS scores of 5 (C) or 6 (D); and populations with (E) and without (F) baseline systemic corticosteroid use. The number at risk at day 27 represents the number of participants with available data at day 28. (G) 60-day all-cause mortality in population 1. The number at risk and number censored before day 28 differ slightly between panels A and G because the day 60 database contained further information on eight participants who were censored at the day 28 database lock but were known to be alive at the day 60 database lock. The number at risk at day 59 represents the number of participants with available data at day 60. For time-to-event endpoints, the p value for baricitinib versus placebo was calculated using an unstratified log-rank test, and HRs and 95% CIs were calculated using a Cox proportional hazards model. The treatment effect was adjusted by all baseline randomisation factors, except when redundant (ie, for baseline corticosteroid use in population 2). (H) Distribution of participants in each NIAID-OS category over time, among patients in the intention-to-treat population with available baseline NIAID-OS scores and at least one post-baseline NIAID-OS score, using last observation carried forward. An NIAID-OS score of 5 represents patients who are hospitalised and require supplemental oxygen, and a score of 6 represents patients who are hospitalised and receiving oxygen support via high-flow oxygen devices or non-invasive ventilation. HR=hazard ratio. NIAID-OS=National Institute of Allergy and Infectious Disease Ordinal Scale.

Figure 328-day all-cause mortality by subgroup
HRs and 95% CIs were calculated with a Cox proportional hazards model. The treatment effect was adjusted by all baseline randomisation factors, except when redundant (eg, for age group [<65 or ≥65 years] in the age subgroup analyses). HR=hazard ratio. NIAID-OS=National Institute of Allergy and Infectious Disease Ordinal Scale. *Participants who, at baseline, required oxygen supplementation and were not receiving dexamethasone or other systemic corticosteroids for the primary study condition.
In population 2, for participants with a baseline NIAID-OS score of 5, 28-day mortality was numerically (but not significantly) lower (HR 0·45 [95% CI 0·13–1·54], nominal p=0·31) in the baricitinib group (four [5%] of 79 patients) than in the placebo group (eight [9%] of 88). In participants with a baseline NIAID-OS score of 6, 28-day mortality was significantly lower (HR 0·20 [95% CI 0·02–1·62], nominal p=0·040) in the baricitinib group (one [6%] of 17) than in the placebo group (eight [38%] of 21).
Table 3Adverse events in the safety population
Data are n (%). Data were assessed from days 1–28.
Discussion
COV-BARRIER is the first international, multicentre, double-blind, randomised, placebo-controlled trial designed to evaluate the potential benefit and safety of baricitinib plus standard of care (which included systemic corticosteroids and remdesivir) for the treatment of hospitalised adults with COVID-19, and is the first to report 60-day outcomes in this population from a double-blind, randomised, controlled trial. This study addresses an important knowledge gap related to the optimisation of treatment strategies for hospitalised patients with COVID-19. In this study, baricitinib plus standard of care (including dexamethasone) did not significantly reduce progression to increased oxygen support or death (the composite primary endpoint) when compared with placebo plus standard of care. However, the group of patients allocated to receive baricitinib did show absolute risk reductions of 5 percentage points in all-cause mortality at 28 days and 4·9 percentage points in all-cause mortality at 60 days, resulting in a number-needed-to-treat of 20 to yield one additional survivor at these two timepoints.
We also report the largest set of randomised, placebo-controlled, safety data from hospitalised patients with COVID-19 treated with an immunomodulatory agent in addition to corticosteroids. The frequencies of treatment-emergent adverse events, serious adverse events, infections, and venous thromboembolic events were similar between the baricitinib and placebo groups, and no new safety signals were detected. Notably, despite COVID-19 being considered a risk factor for thrombosis, treatment with baricitinib was not associated with increased venous thromboembolic events in this setting of short-term use. Additionally, baricitinib in combination with standard of care (predominantly corticosteroids) was not associated with an increase in infections, including serious infections or opportunistic infections, in this hospitalised patient population. This dataset provides clinically relevant safety information for the acute care of these patients in the context of administration of baricitinib with concomitant corticosteroids.
COVID-19 clinical management: living guidance.
,
Coronavirus disease 2019 (COVID-19) treatment guidelines.
were updated following the disclosure of results from the open-label RECOVERY trial in June, 2020,
- Horby P
- Lim WS
- Emberson JR
- et al.
in which a 10·9% relative reduction in mortality was observed with dexamethasone (mortality 22·9%) compared with standard of care alone (25·7%; age-adjusted rate ratio 0·83 [95% CI 0·75–0·93], p
- Horby P
- Lim WS
- Emberson JR
- et al.
By comparison, treatment with baricitinib reduced 28-day all-cause mortality by 38·2% compared with placebo (HR 0·57 [95% CI 0·41–0·78], nominal p=0·0018). The RECOVERY trial also showed a relative risk reduction of 11·4% in 28-day mortality with the anti-IL-6 receptor antibody tocilizumab (31%) versus standard of care (35%; HR 0·85 [95% CI 0·76–0·94], p=0·0028). This benefit was not maintained in the absence of corticosteroid use, but interpretation was limited by small numbers.
Tocilizumab in patients admitted to hospital with COVID-19 (RECOVERY): a randomised, controlled, open-label, platform trial.
In the ACTT-2 trial, the 28-day mortality was 5·1% with baricitinib plus remdesivir and 7·8% with remdesivir alone; however, the study was not powered to detect a difference in mortality between the groups and the use of corticosteroids was limited at baseline and during the trial.
- Kalil AC
- Patterson TF
- Mehta AK
- et al.
In COV-BARRIER, baricitinib plus standard of care showed a 38·2% relative reduction in 28-day mortality compared with placebo plus standard of care and most included patients were being treated with dexamethasone at baseline. Reduction in mortality was also found in our subpopulation of participants who, at baseline, required oxygen supplementation and were not receiving dexamethasone or systemic corticosteroids for the primary study condition. The JAK inhibitors ruxolitinib and tofacitinib have also been associated with reductions in mortality in small, single-country, multicentre, randomised controlled trials.
,
- Guimarães PO
- Quirk D
- Furtado RH
- et al.
To our knowledge, baricitinib showed the largest effect size on mortality of any COVID-19 treatment reported in other randomised trials in hospitalised patients, and showed a benefit in addition to the use of standard of care (corticosteroids) alone.
- Beigel JH
- Tomashek KM
- Dodd LE
- et al.
,
- Kalil AC
- Patterson TF
- Mehta AK
- et al.
,
- Horby P
- Lim WS
- Emberson JR
- et al.
,
Tocilizumab in patients admitted to hospital with COVID-19 (RECOVERY): a randomised, controlled, open-label, platform trial.
The enrolment timeline of COV-BARRIER is also relevant considering the evolving standard of care and the heterogeneity of treatments across different geographical regions. All-cause mortality is the most relevant outcome in trials of patients hospitalised for COVID-19, and baricitinib plus standard of care showed a meaningful reduction in mortality compared with placebo plus standard of care, most notably for participants receiving high-flow oxygen or non-invasive ventilation.
- Sims JT
- Krishnan V
- Chang CY
- et al.
Additionally, the significantly lower mortality by day 60 in the baricitinib group versus the placebo group confirms that the reduction in mortality with baricitinib persists. Future research on baricitinib in patients with COVID-19 should include assessment of the effect of baricitinib at higher doses or with a loading dose to prevent progression events.
- Hasan MJ
- Rabbani R
- Anam AM
- et al.
In summary, our results suggest that baricitinib reduces 28-day and 60-day mortality when used in addition to the current standard of care. The safety profile was similar between the baricitinib group and the placebo group. As such, baricitinib plus standard of care could be a treatment option to reduce overall deaths in the context of the global burden of mortality during the COVID-19 pandemic.
All authors contributed to the concept and design of the trial, data analysis and interpretation, critical revision of the publication, and final approval to submit, and were accountable for the accuracy and integrity of the publication. CEK, RL, SC, and BC have accessed and verified the underlying data in this report.
Declaration of interests
VCM received research grants from the US Centers for Disease Control and Prevention (CDC), Gilead Sciences, the US National Institutes of Health (NIH), Veterans Affairs, and ViiV; received honoraria from Eli Lilly and Company; served as an advisory board member for Eli Lilly and Company and Novartis; and participated as a study section chair for the NIH. AVR received research grants from Eli Lilly and Company; and served as a speaker or consultant for AbbVie, Eli Lilly and Company, Novartis, Pfizer, Roche, Sobi, and Union Chimique Belge. SB, CEK, VK, RL, MLBP, AC, SC, BC, PR, XZ, and DHA are employees and shareholders of Eli Lilly and Company. JDG received research support from Eli Lilly and Company, Regeneron Pharmaceuticals, and Gilead Sciences; grants from Eurofins Viracor and the Biomedical Advanced Research and Development Authority (administered by Merck); speaker fees from Eli Lilly and Company, Gilead Sciences, and Mylan Pharmaceuticals; and advisory board fees from Gilead Sciences. JAA served as a speaker and scientific advisor for AstraZeneca, Boehringer Ingelheim, BMS, Eli Lilly and Company, Foundation Medicine, Novartis, MSD, Roche, and Takeda. VE received a research grant from Eli Lilly and Company. MS received research grants from Eli Lilly and Company, NIAID, and Novartis; and served as a board member for NBOME, Osteopathic Founders Foundation, and COGMED. EWE received research grants from the CDC, NIH, and Veterans Affairs; and served as an unpaid consultant for Eli Lilly and Company. RDP declares no competing interests