GRADE summary of findings: corticosteroids in patients with severe COVID-19, based on indirect evidence from observational studies of patients admitted to hospital with influenza
| Outcomes | Relative effects | Absolute effect estimates | Quality of evidence | Plain language summary | |
|---|---|---|---|---|---|
| Baseline risk for control group, % | Difference (95% CI), % | ||||
| Mortality | OR 1.70 (95% CI 1.31 to 2.21) Based on data from 8530 participants from 11 observational studies43–45,47,52,55,57–59,61,68 | 10.4* | 6.1 (2.8 to 10.0) | Very low (serious indirectness‡) | We are very uncertain of the effect of corticosteroids on mortality |
| Superinfection | OR 2.74 (95% CI 1.51 to 4.95) Based on data from 6114 participants from 7 observational studies43,44,47,52,55,57,65 | 7.2† | 10.3 (3.3 to 20.5) | Very low (serious risk of bias and indirectness§) | We are very uncertain of the effect of corticosteroids on superinfections |
| Mechanical ventilation | OR 5.54 (95% CI 1.83 to 16.80) Based on data from 4364 participants from 4 observational studies52,57,59,61 | 41.8§ | 38.1 (15.0 to 50.6) | Very low (serious risk of bias and indirectness†) | We are very uncertain of the effect of corticosteroids on need for mechanical ventilation |
Note: CI = confidence interval, COVID-19 = coronavirus disease 2019, GRADE = Grading of Recommendations Assessment, Development and Evaluation, OR = odds ratio.
↵* Baseline risk from a study of patients with severe COVID-19 without corticosteroids use.2
↵† Baseline risk comes from the median effect of the control group in the included studies.
↵‡ Observational studies started at low quality of evidence. Additional concern was indirectness (we applied the results to patients with severe COVID-19, but the relative effect was derived from patients admitted to hospital with influenza).
↵§ Observational studies started at low quality of evidence. Additional concerns included high risk of indication bias because unadjusted estimates were included, and indirectness (we applied the results to patients with severe COVID-19, but the relative effect was derived from patients admitted to hospital with influenza).