Dexamethasone, COVID-19 and the RECOVERY trial. St Emlyn’s

Cite this article as: Simon Carley, "Dexamethasone, COVID-19 and the RECOVERY trial. St Emlyn’s," in St.Emlyn's, June 28, 2020,

We have previously covered the RECOVERY trial on St Emlyn’s noting that the first results out of this large pragmatic, adaptive design platform RCT showed no benefit to Hydroxychloroquine.

A few weeks ago a press statement was released that claimed a benefit to dexamethasone in patients requiring oxygen and/or ventilatory support. The release of the data, and instructions to change practice via press release was widely condemned as poor scientific practice in social media and also at local levels within the hospitals we work in. Nobody changes practice on the basis of a press release as there was very little data (just two sentences) in the press release.

Similar criticisms about releasing data via press release were aimed at the trial’s release of data on Hydroxychloroquine, which again has not been seen in print at the time of writing. No explanation has been given by the authors as to why the full data has not been released, leading to many accusations, some reasonable and some quite frankly bizarre (you can search twitter for yourself). It seems to me that they are sorely in need of some social media advice to develop a strategy for knowledge translation in the modern world.

You can listen to our concerns about medicine by press release on the St Emlyn’s podcast here.

Thankfully we now have a pre-print document released to MedRxiv on the dexamethasone arm of the trial1. Whether this was in response to the thousands of requests online is uncertain, but whatever the reason we are delighted to see it.

The abstract. is below, but as always we strongly recommend that you read the full paper and come to your own conclusions

What kind of paper is this?

This is an open-label randomised controlled trial. It is a platform adaptive trial which we described in a previous post, but in essence the broad trial methodology and processes can continue to run as new drugs come in and out to be tested. It’s a fantastic design to use in a pandemic and is something we should look to developing for future pandemics in other countries. Our own Rick Body is co-running a similar platform study for diagnostic tests in COVID-19, this will be known as the CONDOR study.

The RECOVERY Trial is a large, randomised controlled trial of possible treatments for patients that has a consistent design (in this case an open label RCT) but which permits treatments to enter and exit the trial as the become available. Thus far we have been randomising patients to the following treatments as part of RECOVERY.

 Lopinavir-Ritonavir
 Low-dose Dexamethasone NOW STOPPED
 Hydroxychloroquine NOW STOPPED
 Azithromycin
 Tocilizumab
 Convalescent plasma

The use of Tociluzamab and convalescent plasma are in subgroups of patients using the following flowchart, which is a little complex, but will allow us to look at individual and combined therapies.

DEX- dexamethasone, HCQ-hydroxychloroquine, AZM-azithromycin, LOP-lopinovir/ritonivir, SOC-standard of care

Interim analyses are pre-planned and overseen by a data monitoring committee. The investigators remain blinded to the outcomes unless the DMC permits the release of data.

The dose of Dexamethasone was 6mg once daily for up to 10 days.

Tell me about the patients

This is a very pragmatic trial. Essentially any patient admitted to hospital is eligible for the trial. At the start only non-pregnant adults were enrolled, but more recently the trial has been opened to pregnant women and children although they do not feature in these results.

The primary outcome was all cause mortality at 28 days.

What about the results?

2104 were randomised to dexamethasone and 4321 were randomised to usual care. Many patients had co-morbidities such as diabetes 24%, lung disease 27% and heart disease 21%. Most patients were men. The cohort looks similar to other trials in COVID-19 suggesting that it reflects the typical patients we see attending hospital with this disease. Patients were well matched at baseline.

95% of patients randomised to Dex received it and the average treatment length was 6 days.

The results clearly show that there is a benefit to patients admitted to hospital and who require oxygen. Significantly fewer patients allocated to dexamethasone died than in the usual care group (454 of 2104 patients [21.6%] allocated dexamethasone vs. 1065 of 4321 patients [24.6%] allocated usual care; rate ratio, 0.83; 95% confidence interval [CI], 0.74 to 0.92; P<0.001) The magnitude of that effect is greatest in those requiring ventilation. The Cox regression survival curves are shown below.

These equate to some startling Numbers Needed to Treat as calculated by Swami.

So this is still non peer reviewed data, but as someone who has been involved in the delivery of this trial and who has seen how the design has been adopted and adapted for COVID-19 I am confident that the results stand and are suitable for clinical practice.

There are some caveats. The trial is open label and would have been better if blinding had been done. Overall, and although this trial was running in al major UK hospitals, only 15% of patients with COVID-19 were entered into the trial. This is non-sequential recruitment and some selection of suitability, together with research team availability may have an effect on patient selection. The time to randomisation was quite long, presumably owing to a combination of patient disease progression and research team availability. If we now translate the findings to clinical practice it is likely that Dexamethasone will be given earlier, one might argue that this would be even better, but we don’t know that to be true.

The mortality rates in the paper have been criticised in some circles as suggesting that this was a sicker group of patients than normal, but the mortality rates in the ICU and non-ICU patients was comparable to UK practice. What does differ is the definition and practice of ICU care in the UK as compared to the rest of the world (where the term ICU may be more liberally applied), so be cautious of international comparisons of mortality rates.

The full paper deserves your attention as there is lots more in the detail and on the secondary analyses.

Thoughts on communication.

This trial was originally revealed as a press release. This caused a huge amount of scepticism, concern and suspicion amongst the international medical community, and undermined the message of the trial. I sincerely hope that the RECOVERY team look again at their publication strategy and in particular at their social media strategy and engagement which could be greatly improved. Knowledge dissemination during this pandemic needs to adapt to new technologies (they’re not really that new to be honest), and all trialists and research groups need to be agile enough to embrace this. We are still awaiting a pre-print on the data for hydroxychloroquine, the results of which were also released by press release and with only summary data.

Science by press release is not science.

The bottom line

Patients admitted to hospital and who require oxygen should be given dexamethasone.

Further reading

Interestingly, we all come to a fairly robust consensis on this trial.




  1. 1
    Horby P, Lim WS, Emberson J, et al. Effect of Dexamethasone in Hospitalized Patients with COVID-19: Preliminary Report. 2020; published online June 22. DOI:10.1101/2020.06.22.20137273.

Cite this article as: Simon Carley, "Dexamethasone, COVID-19 and the RECOVERY trial. St Emlyn’s," in St.Emlyn's, June 28, 2020,

1 thought on “Dexamethasone, COVID-19 and the RECOVERY trial. St Emlyn’s”

  1. Great study design, perfect example of pandemic EBM approach.
    I wish we could have used it since the very first days of our surge, when the prevalence effect lead to our ED the sickest and oldest patients with COVID19 pneumonia; we started the use of steroids in all O2-dependant patients after the 2nd week of the outbreak.
    We could’ve enrolled more than 1.000 patients in 8 weeks and have some early answers for the rest of the world …
    In case of COVID19 recurrence, would be awesome to share this study design all over the word.

Thanks so much for following. Viva la #FOAMed

Scroll to Top