COVID-19: A Brighter Future for Medical Research?
Posted on 16th July 2020 by Robert Wight
For many, the past few months have been something of a blur. This is certainly true for those trying to keep up with the latest research on the pandemic – a PubMed search for ‘COVID-19’ returns over 29,000 results as of the 7th July. Amongst this deluge of new research, there have been well publicised examples of questionable scientific practice, from so-called ‘science by press release‘ to two of the most high profile retractions in recent memory.
But there have been more positive notes. The pandemic has also highlighted some of the ways in which medical research, and science more generally, can improve its practices in the future. Here we examine two prominent examples: the use of large, simple randomised trials to address clinical uncertainty regarding existing treatments; and the increasing use of preprints (as well as the re-evaluation of traditional peer review) to increase transparency in scientific publishing.
Randomised Controlled Trials
The gold standard study design for testing new medical interventions is the randomised controlled trial (RCT). There are sometimes barriers to conducting suitable RCTs, however, and many commonly used medical treatments have not been thoroughly tested in this way. Some of the barriers to conducting RCTs are not fundamental limitations but avoidable frustrations, arising from inflexible, disproportionate regulation (1,2).
The pandemic has seen RCTs conducted at unprecedented speed. The RECOVERY trial in the United Kingdom, evaluating existing drug treatments for patients hospitalised with COVID-19, enrolled its first patients less than two weeks after it was conceived (3,4). Within 12 weeks, it had been able to recruit over 11,000 participants (live updates of the RECOVERY trial available here) and had reported preliminary results for its hydroxychloroquine and dexamethasone arms. The finding that dexamethasone reduces mortality in seriously ill patients makes it the first drug to have been shown to reduce mortality in COVID-19 (5). There have been some criticisms of certain aspects of RECOVERY’s design and reporting. These are discussed (and addressed by one of the investigators) in a recent feature article in The BMJ (4).
The timeframe in which RECOVERY has produced important results is a testament to what is possible in medical research. In wider medical practice, there are a great number of unanswered questions regarding the use of established treatments. Many of these pertain to common medical conditions and could be answered in a timely fashion with large, simple RCTs. This would require a shift in culture towards greater integration of research into clinical practice (this is well established in paediatric oncology, where nearly 70% of patients are enrolled in clinical trials (6)), as well as a re-appraisal of trial regulation. For some detailed recommendations, this paper by Salman et al is a good place to start: ‘Increasing value and reducing waste in biomedical research regulation and management‘.
Preprints and Peer Review
On 25th May, a number of RCTs studying hydroxychloroquine were temporarily suspended (the RECOVERY trial continued recruitment following independent review by its data monitoring committee). This was because a large observational study published in The Lancet had suggested the drug may have increased the incidence of cardiac arrhythmia in COVID-19 patients. But concerns were soon raised regarding both the paper’s statistical analysis (7) and the integrity of the data it was based on.
On the 4th June, The Lancet retracted the study. The New England Journal of Medicine also retracted another study from the same group of authors. Following these high-profile retractions, many asked how the papers had passed the peer review processes of two of the world’s most prominent medical journals. One answer is that peer review is simply not designed to conduct thorough inspections of studies’ underlying data. But there are also more fundamental concerns over the ability of peer review to ensure the trustworthiness of published research.
Another way that research is scrutinised is through post-publication review, where a larger audience of experts from different disciplines are able to openly discuss the merits of a published study. In recent years, this process has been extended into the pre-publication space, thanks to the growing use of preprints. Preprints are completed papers, posted on dedicated servers such as medRxiv, that have not yet been published in a journal (most journals will accept papers that have previously been posted as preprints, but this is not always the case). Whilst medical science has been relatively slow to adopt preprints, they have been used extensively during COVID-19 as they allow new research to be released (and scrutinised) immediately.
There are understandable concerns that preprints may be used prematurely to guide clinical practice, or that news outlets might broadcast findings to the public which later turn out to be false. The possibility that findings may be over-interpreted or new results not treated with sufficient caution is very real, yet this is also true of published research.
COVID-19 has been an excellent example of the inefficiency that exists in clinical research. Thousands of patients around the world have been treated with unproven experimental drugs, either in observational studies or with no follow up whatsoever. If a fraction of these patients had been enrolled in randomised controlled trials to test those drugs, we would have a far greater collective knowledge base to guide our treatment. Trials such as RECOVERY offer a template for the kind of simple, pragmatic RCTs that could be routinely used to answer important clinical uncertainties about common conditions.
As well as improving the way research is organised and conducted, there is also scope to improve the way it is reported. COVID-19 has seen a rapid rise in the use of preprints and has highlighted the frailties of the current publishing system. It is important to consider alternative reporting practices that may improve the reliability and trustworthiness of scientific findings.