Over the course of three weeks in June, University of Oxford researchers running a closely followed study in the U.K. changed how doctors around the world treat COVID-19.
In short succession, the trial investigators made public results showing a controversial malaria pill and two widely used HIV drugs don't help people hospitalized with the infectious disease. More importantly, in between those two announcements, the Oxford team determined that dexamethasone, a cheap and readily available corticosteroid, helped COVID-19 patients sick enough to need breathing support survive.
It was a dramatic outcome, the first time a medicine had been shown to help reduce the risk of death from COVID-19 for that group of patients in a controlled, randomized clinical trial. The findings were a "much-needed reason to celebrate," said Tedros Adhanom Ghebreyesus, the director-general of the World Health Organization.
RECOVERY, as the trial is known, has now produced a good deal of what the world knows about which medicines work for treating COVID-19 and which don't. Results have led the U.S. Food and Drug Administration to rescind an emergency authorization for hydroxychloroquine, the U.K. government to grant an immediate approval for dexamethasone as a coronavirus treatment and the U.S. National Institutes of Health to recommend the steroid in patients on breathing support.
RECOVERY's findings hold weight because of its design — randomizing patients to either treatment or control groups is one of the surest ways to prove a benefit is real and not the product of chance — and because of the large number of patients enrolled.
Many of the hundreds of clinical trials rapidly set up in response to the coronavirus pandemic can claim neither advantage. A sizable portion are too small; others lack a control or placebo comparison, which are sometimes miscast as slowing patient access to treatments. Such criticisms have been given more weight during past pandemics as well as the current one, most dramatically in the debate that's played out over whether to make hydroxychloroquine widely available before obtaining conclusive proof it actually works.
The Oxford researchers, however, have been able to conduct a high-quality study, while moving at the kind of speed required for addressing an urgent public health crisis. In doing so, the RECOVERY investigators have provided a "model for learning faster," wrote Scott Gottlieb and Mark McClellan, both former FDA Commissioners, in a recent editorial in The Wall Street Journal.
A key component of RECOVERY's success is its simplicity. Investigators, led by Oxford's Peter Horby and Martin Landray, drafted a trial plan that asked physicians in the study to collect only the most critical information about the patients they enrolled, and measure clear outcomes: death, discharge from hospital and breathing support.
"We knew the hospitals would be up against it," said Richard Haynes, a unit lead for RECOVERY and a professor at Oxford, in an interview. "If they were going to do a trial, it would need to be as easy as possible so they could recruit people as part of routine care."
RECOVERY's protocol, a kind of master layout for a clinical trial, was initially just 22 pages and, now, remains a lean 33 pages, less than one-tenth the length of the protocol used by the NIH to run a COVID-19 trial of Gilead's antiviral drug remdesivir.
As a result, physicians at the 176 National Health Service hospitals that participated in RECOVERY were able to quickly enroll patients, some days adding as many as 500 people into the trial, said Thomas Jaki, a co-investigator and professor of statistics at Lancaster University. Since mid-March, when RECOVERY began, more than 12,000 people in the U.K. have taken part, roughly one out of every six patients hospitalized with COVID-19 in the country.
"We got the timing right," said Jaki in an interview. "We started the study just before the large wave of cases started in the U.K."
RECOVERY also benefited from the full-throated support of the U.K. government and the NHS system. It was one of three studies designated as a key national trial, making it easier for hospitals to decide which to join and prioritize.
Speed comes with trade-offs, of course. RECOVERY did not collect any lab or virological data on patients in their study, apart from a diagnosis of SARS-CoV-2 infection. Researchers therefore can't say how different treatments affected different markers of patient's response to the virus.
In a pandemic, however, answering simpler questions about whether a treatment can help people live may matter more, suggested Haynes.
"It's much more useful now to be able to say confidently that if you're hospitalized hydroxychloroquine doesn't benefit you, lopinavir/ritonavir doesn't benefit you, and dexamethasone definitely reduces risk of death if you need oxygen," he said. "If we added in extra bits, we wouldn't have answered those questions by now."
RECOVERY's investigators have been criticized for releasing results without also publishing the supporting data needed to more fully evaluate their findings. Their determination that hydroxychloroquine wasn't effective in treating patients hospitalized with COVID-19 — the strongest sign the much-debated drug is unlikely to play a role in combating the pandemic — was announced June 5. More than a month later, neither a draft manuscript of the data nor a published, peer-reviewed paper are available.
The same is true of the investigators' findings on the combination of lopinavir and ritonavir, two HIV drugs that have been widely studied as potential early treatments for coronavirus infections.
"We've been criticized for doing science by press release. We're definitely well aware of that," said Haynes. But he, as well as Jaki, argued that because the treatments they've tested are readily available and in use, delaying for weeks to go through a peer-review process wouldn't be ethical when an announcement could immediately shape patient care for the better.
That was especially true for dexamethasone, which showed a dramatic 33% reduction in the risk of death among patients on mechanical ventilation. RECOVERY investigators prioritized drafting the paper for those results, publishing a manuscript on the pre-print server medRxiv less than a week after their announcement on the drug.
Papers for hydroxychloroquine and lopinavir/ritonavir are in progress, Jaki said.
RECOVERY, designed to be "adaptive," is also testing the anti-inflammatory drug Actemra, convalescent plasma and the antibiotic azithromycin. So far 800 patients have been randomized to either Actemra or control, and 194 to either plasma or control, investigator Landray recently said on Twitter.
RECOVERY isn't the only large, randomized study to test multiple treatments. Others by the NIH, the WHO and health authorities in France have moved quickly as well.
Earlier this month, the WHO said it would also stop testing hydroxychloroquine and lopinavir/ritonavir after finding no evidence of significant benefit. The NIH, meanwhile, has moved on to testing remdesivir alongside an Eli Lilly pill called Olumiant in their adaptive study, which had earlier found remdesivir shortened time to clinical recovery.
Their progress could similarly prove once again the strength of randomized clinical trials, even in a pandemic.