Update, 23 June 2020: Since this post was first published on 19 June 2020, a pre-print of the Recovery trial was published on MedRxiv. Well done to the authors on making this available in such a timely fashion. It is worth noting that the document is not yet peer-reviewed and may yet change. In Australia, we recommend taking clinical management guidance from the National COVID-19 Clinical Taskforce’s living evidence guideline.
Update, 6 July 2020: Tim Morris and colleagues have written an in-depth statistical review of the Recovery trial and made it available (as a preprint, of course!) via this link: https://zenodo.org/record/3928540. I highly recommend reading this review to lean more about the Recovery trial and clinical trials in general.
On 16 June an exciting announcement about the Recovery trial was made. This press release heralded that dexamethasone has mortality benefits in subgroups of patients with coronavirus disease 2019 (COVID-19). In this press release one of the authors is quoted as saying “dexamethasone should now become standard of care in these patients”. Clearly, evidence of benefit for a cheap, widely available, and likely safe therapeutic agent is most welcome. The expectation that a press release alone should guide clinical practice is not.
I should state that I have equipoise about the role of glucocorticoids in both COVID-19 and in acute respiratory distress syndrome (ARDS). Hough (2014) wrote an excellent summary of our knowledge of the role of steroids in ARDS up until 2014. More recently, the updated 2019 Cochrane review could not determine with certainty whether glucocorticoids were beneficial for ARDS. Prior to this press release, very smart expert clinicians varied in their practice – for both COVID-19 and ARDS in general – based on the uncertainty in the extant literature. This was reflected in the variance in recommendations in guidelines from major bodies, with the Surviving Sepsis Campaign guidelines one of the only major global guidelines supporting a role for glucocorticoids in COVID-19 with ARDS, while acknowledging that this was based on weak evidence. As COVID-19 appears to have a hyper-inflammatory phase, and may be more inflammatory than other causes of ARDS, the appeal of an immunomodulatory therapy is understandable. I can see both sides, hence my equipoise, and that is why further research in this area was vital.
I am sympathetic to the plight of the Recovery trial researchers. As discussed previously, the unprecedented nature of the COVID-19 pandemic has amplified the ever-present tensions that permeate the early adoption of therapies in medicine. As such, they may feel compelled by an ethical imperative to disseminate this knowledge as widely as possible as quickly as possible. It has been argued, by the brilliant Vinay Prasad no less, that based on the press release (as well as the reputation of the researchers involved and the cheap, likely safe nature of the intervention), not adopting dexamethasone may be unethical. I disagree, and think that the “press release” is passing the ethical buck. If one is certain enough of the results to make a press release, one should be able to share a preprint detailing the methods and the results. This would prevent clinicians, guideline advisory committees, and researchers being stuck between a rock and hard place as they try to decide what to do next.
Press release-based medicine is bad practice. We have already been recently burned by the Surgisphere scandal, for instance, which resulted in some of the authors retracting their own publication. There are currently over 20 COVID-19 studies listed by Retraction Watch as retracted or temporarily retracted. Good clinical practice comes from critical thinking and being able to appraise the evidence. We can not do this adequately without access to the full methods and results of the study.
The advantage of a preprint is that not only do you and I get to critically appraise the methods and results, so does everyone else in the critical care and research communities (including many people much smarter than me). The benefits of multiple perspectives through this global peer review are enormous. The Surgisphere debacle, for instance, was brought to a head by readers identifying issues with the data that had not been picked up by the journal or their peer reviewers. In no way do I claim or expect this to happen with the Recovery trial, yet we are playing with fire if we start thinking that any researcher is immune to error. Not that we should take a preprint as gospel, just that it is vastly superior to a press release for informing clinicians, guidelines, and researchers.
I wonder, what are the barriers to a preprint being released? Is it due to a journal embargo? I hope not – I’m sure a journal could be found that would still publish the article after a preprint, and which journal that is hardly matters to patients; the article will be widely read and cited regardless. Is it fear that mistakes or errors will be identified? Is it something else? If there is enough confidence and enough known to make a press release, I don’t understand why a preprint containing all available methods and results cannot also be released.
Some argue that as the research protocol is already published online, the press release is enough to guide practice. Prior publication of a protocol is routine for a high quality trial, but does not replace the methods section of a research article. The methods section can often be truncated by referring to the published protocol but there may be differences and explanations of any changes or decisions. The methods describe what was actually done, not what was planned. Were there any protocol violations? Even more critically, we need the full results to make informed decisions. What are the full characteristics of the patients included in the study and those that responded to therapy? Were there adverse effects? Is the timing of the intervention important? In countries like Australia and New Zealand, where ventilated ICU patients with COVID-19 likely already have outcomes as good or better than the Recovery trial participants who received dexamethasone, the decision to implement a new therapy is non-trivial. Do the results apply to our patients if they already have better outcomes?
The press release provides a dilemma for other researchers too. What are the implications for other trials currently in progress studying glucocorticoid steroids in patients with COVID-19? Should they stop recruiting based on a press release? Widespread adoption of dexamethasone in clinical practice may prevent further studies addressing remaining questions from being performed. Furthermore, use of dexamethasone in the wrong patients could lead to shortages for other patients that would benefit (including select COVID-19 patients and patients with a host of other conditions from croup to various forms of lymphoma to autoimmune diseases).
Some argue that the researchers and the organisations involved are reputable and have excellent track records. I do not doubt this and I agree that it is reassuring. However, science is about being skeptical, not blindly following or listening to handed down wisdom. Importantly, anyone can make mistakes – as we have seen with the numerous retractions and corrections of COVID-19 research publications so far.
On the one hand, I congratulate the researchers on completing this important study and I look forward to reading what I hope will be a landmark study. On the other, I maintain that science by press release is bad practice and continues to set damaging precedents for the guidance of patient care.
novel coronavirus of COVID-19