Systemic Corticosteroids for COVID-19

Systemic corticosteroids probably reduce all-cause mortality in hospitalized COVID-19 patients

Benefits in NNT

1 in 33 may be helped (death prevented)
May lower the risk of death by 3%, please see the Caveat section for explanation of uncertainty surrounding the benefit effect

Harms in NNT

Not reported
Not reported
View As:


Wagner C, Griesel M, Mikolajewska A, et al. Systemic corticosteroids for the treatment of COVID-19. Cochrane Database Syst Rev. 2021;2021(8):CD014963.

Study Population: 8,075 participants in 11 randomized controlled trials that evaluated systemic corticosteroids for people with COVID-19, irrespective of disease severity

Efficacy Endpoints

All-cause mortality (at longest follow-up available); ventilator-free days; new need for invasive ventilation

Harm Endpoints

Adverse events; hospital-acquired infections


The novel coronavirus-19 emerged in late 2019 and has resulted in a global pandemic with over 5.2 million deaths to date.1 A systemic inflammatory response with excessive release of cytokines and other inflammatory mediators can lead to lung injury and low oxygen levels (hypoxia).2 The RECOVERY trial, the world's largest clinical trial evaluating treatments of COVID-19, was conducted in 185 sites in the United Kingdom. It was the first trial to suggest that systemic corticosteroids, such as dexamethasone, may decrease the risk of death and decrease the progression of disease in nonventilated patients (from ventilator-free to requiring invasive mechanical ventilation).3 Since publication of this trial, several other trials have explored the use of corticosteroids in COVID-19.

The Cochrane systematic review and meta-analysis discussed here evaluates the effect of systemic corticosteroids in COVID-19 patients.4 It included 11 randomized controlled trials (RCTs) comprising 8,075 participants diagnosed with COVID-19. Of these participants, 3,072 were randomized to receive systemic corticosteroid plus standard care and the remainder received standard care alone. A majority of patients receiving corticosteroids were treated with dexamethasone (n = 2,322). The outcomes of interest included all-cause mortality, ventilator-free days, a new need for invasive mechanical ventilation, serious adverse events, all adverse events, and hospital-acquired infections.

The meta-analysis found that all-cause mortality was decreased in the systemic corticosteroids plus standard care arm compared to standard care alone (relative risk [RR] = 0.89, 95% confidence interval [CI] = 0.8 to 1.00, absolute risk difference [ARD] = 3%, NNT = 33, nine RCTs, n = 7,930, median follow-up = 28 days, quality of evidence = moderate).

Only one RCT showed that the number of ventilator-free days was larger in corticosteroid arms (mean difference = 2.6, 95% CI = 0.67 to 3.43, quality of evidence = low). There was not enough evidence to determine whether corticosteroids affected the need for invasive mechanical ventilation, liberation from mechanical ventilation, quality-of-life measures, viral clearance, and need for dialysis.

Only two of the included RCTs measured serious adverse events and only five RCTs reported the risk of hospital-acquired infections. Because of the high risk of bias, heterogeneous definitions, and underreporting, the authors of the meta-analysis were uncertain about the size and direction of the effect when assessing harm outcomes.


Given that the upper limit of the 95% CI for the reported relative risk of death (all-cause mortality) reaches 1 (possibility of equal risk in treatment and placebo groups), the survival benefit form systemic corticosteroids is not clearly established. This explains why we labeled this benefit “probable.” Future high-quality trials could swing the pendulum toward more robust evidence of benefit or lack thereof.

There is no universal definition of “standard therapy” for the treatment of COVID-19, so patients in the different studies received a variety of other drugs such as antibiotics, hydroxychloroquine, convalescent plasma, and other treatments that might have confounded the results.

A subgroup analysis in the RECOVERY trial suggested that dexamethasone, when given to nonventilated patients, reduced the progression to invasive mechanical ventilation. However, the Cochrane meta-analysis did not reach the same conclusion. Moreover, because none of the trials enrolled asymptomatic, mildly ill patients, or those who were not hospitalized, there was not enough evidence to conclude that corticosteroids decreased the progression of mild or moderate COVID-19 disease to severe disease.

Corticosteroids have potential side effects such as immunosuppression, increased risk of infection, hyperglycemia, increased risk of gastrointestinal hemorrhage, and impaired wound healing.5 However, most of the adverse events from corticosteroids are associated with chronic use. Regardless, most of the included trials either did not measure or did not report adverse events. It is thus not possible to comment on the magnitude of harm associated with this treatment in COVID-19 patients.

A subgroup analysis from the RECOVERY trial suggested that mortality was higher in the subgroup of patients not on oxygen who received dexamethasone. This difference was not statistically significant. Similarly, the Cochrane meta-analysis also performed a subgroup analysis that stratified patients based on the need for respiratory support at randomization. This subgroup analysis also showed a negative effect on mortality from administration of dexamethasone in the participants who did not require respiratory support. The difference was also not statistically significant. The findings of such subgroup analyses simply generate hypothesis for future trials and drawing conclusions from them is not appropriate.

Over 80% of the study participants were from high-income countries and there were no studies from low-income countries, which may limit the generalizability of these results. Lower-income countries may have limited hospital resources, such as hospital beds, supplies of medications and oxygen, equipment for respiratory support, and medical personnel, and therefore may not experience the same benefits of systemic corticosteroid treatment delivered in well-resourced hospital settings.

Since most of the studies included in the meta-analysis used dexamethasone, the benefits may not be generalizable to all systemic corticosteroids. Only one study compared different types of systemic corticosteroids (methylprednisolone vs. dexamethasone). There was no analysis comparing different doses (range = 150 to 5,000 mg daily hydrocortisone equivalents) or duration of corticosteroids (range = 0 to 20 days). It is thus unclear whether other systemic corticosteroids have the same impact on mortality as dexamethasone or if there is an ideal dose or duration of treatment. Dozens of ongoing and recently completed trials will likely generate additional evidence to shed light on these issues as well as additional data on safety, long-term survival, and other patient-centered outcomes.

All the trials included in the Cochrane systematic review enrolled patients from the beginning of the pandemic and prior to the widespread availability of vaccinations and the emergence of variants such as the delta variant, which has been shown to have increased transmissibility and decreased host immunity.6 It is thus unclear whether systemic corticosteroids demonstrate the same benefits in vaccinated patients, in those who are infected with different variants, or in those with breakthrough infections despite vaccination or prior infection.

Some uncertainty remains in regard to the survival benefit of systemic corticosteroids in COVID patients. In addition, while harms are unlikely from a short course of corticosteroids in patients with severe COVID (the current common practice), we are concerned about the lack of proper tracking and reporting of the harms. Therefore, we have assigned a color recommendation of yellow (unclear if benefits) for systemic corticosteroids in hospitalized COVID-19 patients.

The original manuscript was published in Academic Emergency Medicine as part of the partnership between and AEM.


Kenneth Lu, MD; Richard W. Leno, MD; Selwena R. Brewster, MD
Supervising Editors: Kabir Yadav, MD


December 28, 2021