Randomized Controlled Trial
Study population and setting
This paper, which was posted as a preprint and had not yet been peer-reviewed, reported on 6,425 patients admitted to 176 National Health Service hospitals in the UK with either suspected or laboratory confirmed SARS-CoV-2 between March 19 and June 8, 2020. Exclusion criteria included medical history that might place a patient at “significant risk” in the opinion of the attending clinician. Initially, patients were required to be at least 18 years of age, but this criterion was later dropped. Of the participating patents, 2104 were randomly allocated to receive dexamethasone (six mg/day for up to 10 days) plus standard of care and compared to 4321 patients who received standard of care alone. The primary end point was death from any cause (i.e., all-cause mortality) within 28 days of randomization. Patients who were discharged alive, but not followed for the full 28 days to assess for death because the study had ended, were assumed to have survived for 28 days.
Summary of Main Findings
Dexamethasone reduced overall all-cause mortality by 17% (Risk ratio (RR) = 0.83, 95% CI: 0.75-0.92), with the effect being greatest in patients who had most severe respiratory compromise. For example, dexamethasone decreased mortality by 35% in patients requiring intubation and mechanical ventilation and by 20% in patients requiring supplemental oxygen therapy. (RR=0.65, 95% CI: 0.51–0.82 and RR=0.80, 95% CI: 0.70–0.92, respectively). However, dexamethasone did not affect the mortality of patients not requiring respiratory support (RR=1.22, 95% CI: 0.93–1.61). Specific sub-groups that benefitted from dexamethasone were persons under the age of 70, men, and persons whose symptoms had been ongoing for more than seven days at the time of randomization. Another important concern is that persons who were moderately ill, and did not require oxygen at baseline, may have had slightly worse outcomes with dexamethasone, although this was not statistically significant.
This study benefited from a large, multi-center, prospective randomized treatment population with a comparable control group. The primary outcome measurement was 28-day, all-cause mortality. The team only considered outcomes in people who had either died prior to 28 days or who had a full 28 days of follow-up, avoiding the censoring of data from people who had incomplete follow-up. Follow-up data was available for 95% of patients. A majority (82%) of patients were laboratory-confirmed for SARS-CoV-2, and they were well-balanced between the group that received dexamethasone and the comparator arm. Other investigational agents (e.g., lopinavir-ritonavir, hydroxychloroquine) were infrequently used and if they were used, well-balanced between the two arms.
The manuscript had not yet been peer-reviewed at the time of our review. There was no breakdown by race or ethnicity, which may be a significant predictive risk for worse outcomes. Pre-existing hypertension and obesity, two major risk factors for COVID-19 severity, were also not reported. Clinician opinion on “high risk” as a contraindication to participation in the trial was vague and thus, the results may not be generalizable to populations at “high risk”, however that was defined. Similarly, patients on mechanical ventilation were 10 years younger than those not receiving respiratory support, which may indicate that patients who were more likely to survive were prioritized to receive ventilators; thus, this may limit generalizability of these results to populations requiring respiratory support but with low probabilities of survival. Furthermore, not all patients had RT-PCR confirmed SARS-CoV-2, which may limit the generalizability of these results to populations where the suspected cases outnumber laboratory-confirmed cases. It is also worth noting that this was an open-label study, where the researchers and patients in the study knew who was receiving which treatment. While a masked trial, where researchers and patients do not know who is receiving which treatment, are generally known to be less biased, it is difficult to see how the unmasked approach here would have biased mortality data. Lastly, the study protocol stated that patients randomized to receive dexamethasone could receive it for up to 10 days, and the breakdown in Table S1 shows that the duration of treatment with dexamethasone varied widely across patients [interquartile range: 3-10 days]. Thus, a subgroup analysis conducted on the duration of dexamethasone would be helpful.
This is the largest randomized controlled trial of a pharmaceutical intervention for COVID-19 to date. In addition, the mortality benefit from dexamethasone is the most substantial that has been seen in a therapeutic RCT for COVID-19. The number of mechanically ventilated patients who would need to be treated with dexamethasone is eight to save one life.
This review was posted on: 4 July 2020