New England Journal of Medicine February 25, 2021

Dexamethasone in Hospitalized Patients with Covid-19 (RECOVERY)

RECOVERY Collaborative Group (Peter Horby, Martin J. Landray et al.)

Bottom Line

In hospitalized patients with COVID-19, dexamethasone significantly reduced 28-day mortality among those requiring invasive mechanical ventilation or supplemental oxygen, but showed no benefit and possible harm in patients not requiring respiratory support.

Key Findings

1. Overall 28-day mortality was significantly lower in the dexamethasone group compared to the usual care group (22.9% vs. 25.7%; rate ratio [RR] 0.83, 95% CI 0.75-0.93, P<0.001).
2. The mortality benefit was most pronounced among critically ill patients receiving invasive mechanical ventilation at randomization (29.3% vs. 41.4%; RR 0.64, 95% CI 0.51-0.81).
3. A significant mortality benefit was also observed in patients receiving supplemental oxygen without invasive mechanical ventilation (23.3% vs. 26.2%; RR 0.82, 95% CI 0.72-0.94).
4. No survival benefit was seen in patients who were not receiving any respiratory support at randomization, with a trend toward increased mortality (17.8% vs. 14.0%; RR 1.19, 95% CI 0.92-1.55).
5. Dexamethasone treatment was associated with a shorter duration of hospitalization (median 12 days vs. 13 days) and a greater probability of discharge alive within 28 days (rate ratio 1.10).

Study Design

Design
RCT
Open-Label
Sample
6,425
Patients
Duration
28 days
Median
Setting
Multicenter, UK
Population Hospitalized patients with clinically suspected or laboratory-confirmed SARS-CoV-2 infection (COVID-19).
Intervention Oral or intravenous dexamethasone at a dose of 6 mg once daily for up to 10 days (or until hospital discharge, if sooner).
Comparator Usual standard of care alone.
Outcome All-cause mortality at 28 days after randomization.

Study Limitations

The trial was open-label, though the primary outcome of 28-day all-cause mortality is objective and highly resistant to observer bias.
Long-term outcomes and safety follow-up beyond 28 days were not included in this primary report.
The streamlined, pragmatic design of the trial meant that detailed clinical data—such as laboratory inflammatory markers or specific radiologic findings—were not comprehensively collected.
The rapidly evolving pandemic environment meant that background standards of care varied across sites and changed over the course of the enrollment period.

Clinical Significance

The RECOVERY trial was the first massive randomized clinical trial to definitively identify a life-saving therapy for severe COVID-19. It instantly altered the global standard of care by demonstrating that an inexpensive and widely available corticosteroid drastically reduces mortality in patients requiring respiratory support. Crucially, it also established that dexamethasone should be avoided in milder, non-hypoxic cases where early viral replication predominates, as it showed no benefit and potential harm.

Historical Context

During the early months of the COVID-19 pandemic, the use of systemic corticosteroids was highly controversial. Drawing on experiences with SARS, MERS, and severe influenza—where steroids were linked to delayed viral clearance and secondary infections—major guidelines (including early WHO recommendations) initially advised against their routine use in COVID-19. The RECOVERY adaptive platform trial, leveraging the UK's National Health Service, challenged this paradigm by quickly enrolling thousands of patients and unequivocally proving that modulating the host hyperinflammatory response saves lives in late-stage disease.

Guided Discussion

High-yield insights from every perspective

Med Student
Medical Student

Based on the pathophysiology of COVID-19, why did dexamethasone reduce mortality in patients requiring supplemental oxygen or mechanical ventilation but show a trend toward harm in patients not requiring respiratory support?

Key Response

This highlights the biphasic nature of COVID-19. Early in the disease (mild symptoms, no oxygen requirement), the pathology is driven by active viral replication; giving an immunosuppressant like dexamethasone can hinder the host's antiviral response, leading to worse outcomes. Later in the disease (requiring oxygen/ventilation), the pathology is driven by a hyperinflammatory 'cytokine storm' and ARDS; here, the immunosuppressive effects of dexamethasone blunt the damaging systemic inflammation, improving survival.

Resident
Resident

A patient is admitted to the floor with COVID-19 pneumonia on room air with an SpO2 of 95%, but complains of severe myalgias and high fevers. Do you start dexamethasone based on the RECOVERY trial, and what are the potential clinical consequences of your decision?

Key Response

No, you should not start dexamethasone. The RECOVERY trial demonstrated no benefit and possible harm (rate ratio 1.19 for mortality) in patients not requiring oxygen. Prescribing steroids in this scenario represents harmful overtreatment that can prolong viral shedding, increase the risk of secondary infections, and cause uncontrolled hyperglycemia, without offering any mortality benefit.

Fellow
Fellow

The RECOVERY trial used a dexamethasone dose of 6 mg daily. How does this dose compare in glucocorticoid equivalents to doses typically studied in non-COVID ARDS (such as in the DEXA-ARDS trial), and what is the physiologic rationale for choosing this specific moderate dose?

Key Response

Dexamethasone 6 mg is equivalent to roughly 32 mg of methylprednisolone, which is a moderate dose. Older ARDS trials like DEXA-ARDS often used higher doses (e.g., 20 mg daily for 5 days, then 10 mg). The moderate dose in COVID-19 aims to strike a delicate balance: sufficiently dampening the dysregulated inflammatory response driving lung injury while avoiding profound immunosuppression that could precipitate opportunistic infections like COVID-19-associated pulmonary aspergillosis (CAPA) or severe viremia.

Attending
Attending

The RECOVERY trial was an open-label, pragmatic platform trial. When teaching on rounds, how do you explain why the lack of blinding does not critically undermine the validity of the primary outcome (28-day mortality), while it might affect secondary endpoints?

Key Response

A key teaching point in evidence-based medicine is matching trial design to the endpoint. Mortality is an objective, unequivocal, 'hard' endpoint that is highly resistant to observer bias, meaning the lack of blinding is unlikely to skew the primary result. However, subjective secondary endpoints like the decision to intubate or clinical symptom scores could be influenced by the physician knowing the patient's treatment allocation, which is a vital nuance when interpreting open-label data.

Scholarly Review

Critical appraisal through the lens of expert reviewers and guideline development

PhD
PhD

The RECOVERY trial utilized an adaptive platform design with a master protocol. What are the statistical advantages and potential multiplicity pitfalls of this design when simultaneously evaluating multiple investigational arms against a shared standard-of-care control group?

Key Response

Adaptive platform trials offer immense statistical efficiency by using a shared control group (reducing total sample size) and allowing for response-adaptive randomization to drop futile arms or graduate successful ones quickly. However, the methodology requires rigorous statistical safeguards to control the family-wise error rate across multiple hypotheses. Furthermore, temporal drift in the 'usual care' group (as standards of care evolve during a pandemic) requires complex modeling to ensure concurrent controls are properly weighted against specific interventions.

Journal Editor
Journal Editor

In the RECOVERY trial, approximately 17% of patients allocated to the dexamethasone arm did not receive the drug, and 8% in the usual care arm received systemic corticosteroids. How does this degree of non-adherence and crossover affect the interpretation of the intention-to-treat (ITT) analysis, and what supplementary analyses would a rigorous peer reviewer demand?

Key Response

Crossover and non-adherence in an open-label trial dilute the treatment effect, biasing the ITT analysis toward the null hypothesis. Because RECOVERY still found a significant mortality benefit despite this, the ITT result is considered a conservative estimate. As an editor, one would demand a per-protocol or 'as-treated' sensitivity analysis to ascertain the true magnitude of the biological efficacy of dexamethasone, ensuring that the crossover did not obscure even greater benefits or unique harm signals.

Guideline Committee
Guideline Committee

Following the RECOVERY trial, panels like the NIH and WHO rapidly issued strong recommendations for dexamethasone in severe COVID-19 but strong recommendations against it in non-hypoxemic patients. How does a single pragmatic trial justify a 'strong' recommendation under GRADE methodology, and how should guidelines address the subgroup interaction?

Key Response

Under GRADE methodology, while strong recommendations typically rely on multiple RCTs, a single trial can suffice if it is large, methodologically sound, and demonstrates a large magnitude of effect on a critical outcome like mortality with high certainty. The distinct subgroup interaction in RECOVERY (benefit in hypoxemia, harm in normoxemia) provided clear, high-certainty evidence of differing risk-benefit profiles. Current NIH guidelines strictly reflect this by grading dexamethasone as an A-I recommendation for patients on oxygen/ventilation, but explicitly issuing an A-I recommendation against its use in patients not requiring supplemental oxygen.

Clinical Landscape

Noteworthy Related Trials

2020

ACTT-1 Trial

n = 1,062 · NEJM

Tested

Remdesivir

Population

Hospitalized adults with Covid-19

Comparator

Placebo

Endpoint

Time to recovery

Key result: Remdesivir was superior to placebo in shortening the time to recovery in hospitalized adults with Covid-19.
2020

CoDEX Trial

n = 299 · JAMA

Tested

Intravenous dexamethasone

Population

Patients with moderate or severe ARDS due to COVID-19

Comparator

Standard care

Endpoint

Ventilator-free days at 28 days

Key result: Intravenous dexamethasone significantly increased the number of ventilator-free days over 28 days compared to standard care alone.
2020

SOLIDARITY Trial

n = 11,266 · NEJM

Tested

Repurposed antivirals

Population

Hospitalized adults with Covid-19

Comparator

Standard care

Endpoint

In-hospital mortality

Key result: None of the repurposed antiviral drugs evaluated definitely reduced mortality, initiation of ventilation, or hospital stay duration.

Tailored to your role

Want this tailored to you?

Add your specialty or training stage to get role-specific takeaways and more questions.

Personalize this analysis