The Recovery trial is a comparative trial, launched in the UK in March 2020, to evaluate several drugs in patients hospitalised for covid-19. Detailed preliminary results from this trial, concerning the evaluation of systemic corticosteroid therapy, have been released prior to their publication in the scientific press.

The trial was randomised, meaning that these hospitalised patients were randomly assigned to the various groups. There were five groups initially: a control group of patients who received “usual” care, and four groups of patients who, in addition, received one of the four treatments under evaluation: dexamethasone (Dectancyl° or other brands), lopinavir + ritonavir (Kaletra° or other brands), hydroxychloroquine (Plaquenil°), azithromycine (Zithromax° or other brands). The trial was funded by public organisations and private foundations (the Wellcome Trust, and the Bill and Melinda Gates Foundation).

The trial was not blinded, meaning that patients and their healthcare professionals knew which treatment they had been allocated. 176 British public-sector hospitals participated in the trial. Patients were enrolled into the trial on their admission to hospital for Sars-CoV-2 infection during the epidemic.

The detailed results of the evaluation of dexamethasone were released on 23 June 2020. According to a press release, the trial’s evaluation of hydroxychloroquine was halted prematurely due to lack of efficacy (28-day mortality of about 25.7%, versus 23.5% with usual care). The trial is still in progress for the other groups.

The protocol-planned treatment for the corticosteroid group was 6 mg of oral or intravenous dexamethasone once daily for up to 10 days. Pregnant or breast-feeding women instead received 40 mg of oral prednisolone once daily, or 80 mg of intravenous hydrocortisone twice daily. The primary endpoint was 28-day mortality. For patients whose follow-up period was shorter than 28 days, their vital status on 8 June 2020 was used instead.

In the control group of 4321 patients receiving usual care, dexamethasone formed part of the treatment received in 7% of cases. All 2104 patients in the dexamethasone group received dexamethasone in addition to usual care. Patients were randomly assigned to their group on average 8 to 9 days after their symptoms began. The groups had similar profiles in terms of major prior disease. 64% were men. Through chance alone, the patients in the dexamethasone group were on average about 1 year older than the controls: 67 years versus 66 years. This age difference was taken into account in the statistical analyses. Half of the patients took dexamethasone for fewer than 6 days. Aside from age, more than half of the patients’ medical history included factors known to correlate with poor covid-19 outcomes.

28 days after enrolment into the trial, 21.6% of the patients in the dexamethasone group had died versus 24.6% in the control group (p<0.001), with full follow-up available for 95% of the patients. The difference was particularly marked in the subgroup of patients who received mechanical ventilation: 29% of the patients in the dexamethasone group died, versus 40.7% in the control group (p<0.001). In other words, dexamethasone prevented about 12 deaths in this subgroup for every 100 patients treated. The difference was smaller for patients on non-invasive oxygen therapy: 21.5% died versus 25.0% of the controls (p=0.002), i.e. about 3 deaths were prevented for every 100 patients treated. Dexamethasone appeared ineffective in patients whose condition did not warrant supplemental oxygen, and the results do not rule out the possibility that it actually increased their risk of death: 17.0% of the patients in the dexamethasone group died versus 13.2% of the controls. The difference did not reach statistical significance but constitutes a strong safety signal. The evidence provided by these subgroup analyses is weakened by the fact that they were performed post hoc, without knowing the characteristics of the patients in each subgroup.

The scientists who designed the trial chose dexamethasone because it carries a lower risk of sodium and water retention than other corticosteroids. The other main adverse effects known to occur with moderate doses of corticosteroids in the emergency setting are hyperglycaemia and mental disorders. The information published on 23 June does not specify the serious adverse effects observed during the Recovery trial. The protocol did not call for systematic collection of non-serious adverse effects.

In practice, when patients hospitalised for covid-19 receive oxygen or mechanical ventilation, it is useful to also treat them with 6 mg per day of systemic dexamethasone for up to 10 days. If dexamethasone is unavailable, prednisolone (40 mg per day) is an option with probably similar effects. But in the absence of breathing difficulties warranting oxygen, the addition of corticosteroid therapy appears unnecessarily dangerous.

©Prescrire 24 June 2020

Sources:

• "Covid-19 et troubles respiratoires graves : dans un essai, mortalité réduite par la dexaméthasone" Application Prescrire 24 June 2020.
• "Randomised evaluation of COVID-19 therapy (RECOVERY) - Protocol" (V6.0) 14 May 2020 > HERE
• "Effect of dexamethasone in hospitalized patients with COVID-19: preliminary report" 22 June 2020 > HERE
• "No clinical benefit from use of hydroxychloroquine in hospitalised patients with COVID-19" Statement from the chief investigators of the randomised evaluation of covid-19 therapy (RECOVERY) trial on hydroxychloroquine, 5 June 2020 > HERE

For more information:

• "Covid-19: Follow Prescrire's independent, evidence-based analysis of the pandemic" > HERE