A study suggesting that the malaria drug hydroxychloroquine is dangerous to people with severe cases of COVID-19 is sowing confusion among researchers — and halting clinical trials that are crucial to finding out whether the medication is effective at treating people infected with the novel coronavirus.
In light of the study, this week, the World Health Organization (WHO) has paused enrolment in its trial of the drug as a treatment for COVID-19, as have regulators in the United Kingdom, France and Australia, where similar trials are under way. But the picture isn’t clear-cut: on 28 May, 120 researchers signed a letter to The Lancet, which published the study1, highlighting concerns about the quality of the data and its analysis.
Hydroxychloroquine has had a controversial history during the pandemic, with politicians such as US president Donald Trump endorsing the medication, which is cheap and easy to administer, as a treatment despite scant evidence that it works. Researchers have been eagerly awaiting results from clinical trials to yield a clear answer. But now they fear that the Lancet study, and the negative press coverage that followed, might dissuade patients from joining the trials, which would make it even harder to determine whether the drug works against COVID-19.
“There’s so much swirling around it, people won’t want to enter those trials,” says David Smith, an infectious-disease specialist at the University of California in San Diego. “In which case, it will be an open question that won’t get answered.”
There is currently only one treatment that has shown promise against COVID-19 in a rigorous clinical trial and that drug, remdesivir, is in short supply and must be administered intravenously over several days, typically in a hospital. In the early days of the pandemic, which has now killed more than 350,000 people, many were hopeful that hydroxychloroquine, a relatively cheap and widely available drug, might offer some benefits.
Early laboratory studies suggested that the drug, as well as a similar medicine named chloroquine, might interfere with replication of the coronavirus, but trials in humans have been inconclusive so far. In the meantime, advocacy by politicians such as Trump — who has not only endorsed hydroxychloroquine’s use but said that he has taken it prophylactically — has muddied public perception of the drug. “Some people come in and say, ‘The president takes it, I need to take it,’” says Smith. “And others say, ‘The president’s taking it, I’m definitely not taking it.’”
Researchers have been waiting for results from a number of randomized, controlled clinical trials testing hydroxychloroquine as a COVID-19 treatment.
The Lancet study was not based on such trials. Instead, it was an observational study cataloguing the effects of hydroxychloroquine in about 96,000 patients hospitalized with COVID-19 around the world. The study found the drug had no benefit, and instead found a worryingly higher mortality rate among those who received it: 18%, compared with 9% in those who did not.
Data from observational studies can be useful to get a sense of how a drug performs outside the confines of a clinical trial, in which patients are often more carefully selected for treatment and monitored than in the real world. But observational studies can also be prone to real-world bias: one concern, for example, is that physicians might be more likely to give a drug to patients who are sicker and more likely to die.
Nevertheless, the results have led regulators and clinical-trial sponsors to take action. The WHO has paused enrolment in the hydroxychloroquine treatment group of its large Solidarity trial, an effort in 35 countries to test potential COVID-19 therapies that is also testing three other potential COVID-19 treatments. The UK Medicines and Healthcare Regulatory Agency (MHRA) has similarly paused most of the country’s hydroxychloroquine trials for COVID-19, pending further safety assessments. And in France, where clinical-trial enrolment has been halted, doctors have been forbidden to prescribe the drug outside of clinical trials to people with COVID-19.
The Lancet study’s lead author, Mandeep Mehra, a cardiologist at Brigham and Women’s Hospital in Boston, Massachusetts, says that he agrees with the WHO’s decision to take a hard look at the safety data from the Solidarity trial. But he does not support ending clinical trials of hydroxychloroquine altogether. “We never called for halting clinical trials,” he says. “In fact, we called for the opposite.”
Some of these impacts on clinical trials might not be permanent, says cardiologist Joseph Cheriyan of the University of Cambridge, UK. Enrolments in clinical trials might resume after regulators have been assured that the data from these trials do not indicate that participants are being harmed by hydroxychloroquine. Within a day of the Lancet paper’s publication, the UK’s RECOVERY trial, which has enrolled 10,000 patients to test six potential COVID-19 treatments, consulted its safety board. The board analysed the trial’s interim data and found no sign of harm to study participants from hydroxychloroquine. As a result, the trial can continue enrolling.
Other trials might be able to continue recruiting if safety measures are adopted, but some researchers have found that these can end a clinical trial. In March, rheumatologist Jon Giles at Columbia University in New York City launched a trial determine whether hydroxychloroquine could lessen infection risk within households in which a member had contracted COVID-19. But on 24 April, the US Food and Drug Administration issued a warning about the effects of hydroxychloroquine on the heart, and the ethics committee overseeing Giles’s trial demanded that potential participants have an electrocardiogram first.
Giles’s team didn’t think it was safe to send people to a doctor’s office in the middle of a pandemic to have the measurement taken. “I had to shutter my trial,” Giles says.
At the moment, hydroxychloroquine is nearly the only drug being evaluated in COVID-19 prevention studies such as Giles’s trial, says Cheriyan, who is also carrying out a prevention trial. Cheriyan worries that negative publicity surrounding hydroxychloroquine in hospitalized COVID-19 patients could affect enrolment in his trial among healthy people, even though the studies suggesting possible harms from hydroxychloroquine, which has been used for 60 years to treat other ailments, were in people who were already sick or hospitalized with COVID-19.
In the meantime, researchers are racing against the clock. As the initial waves of the pandemic begin to recede in many countries, the pool of potential trial participants is drying up. This, some worry, means that physicians will not have the medicines they need to better deal with subsequent surges of COVID-19. “If there is a second wave, we need to be better prepared,” says Cheriyan. “We can’t let it affect us the way we have this time.”
In the midst of it all, the Lancet analysis itself has come under fire. Some researchers have pointed to oddities in the data, such as smoking and obesity rates that were reported as nearly equal across all parts of the world studied in the analysis. And because the work relied on proprietary data gathered from medical centres, the raw data were not made available to other researchers.
Following these criticisms, Mehra’s team tracked down some errors — the table reporting smoking and obesity rates, for example, erroneously listed modelled data rather than raw data. The Lancet will soon issue a correction, Mehra says, but the errors did not affect the paper’s conclusions. And the company that gathered the proprietary data, Surgisphere Corporation of Chicago, Illinois, says that it will accept proposals from other researchers to analyse its data.
But some remain sceptical about the study’s conclusion that hydroxychloroquine is a safety risk to some people with COVID-19. Smith says that he and his collaborators have analysed the data that are available and concluded that patients who received hydroxychloroquine in the study were often sicker in the first place than those who did not. Daniel Freilich of Bassett Medical Center in Cooperstown, New York, says that he has independently analysed the data and come to the same conclusion.
But Mehra says that his team accounted for those differences and still found a small difference in mortality. Furthermore, he says, his study merely highlights an association between the treatment and increased death, but because it is an observational study and not a clinical trial, it cannot exclude the possibility that other factors may have contributed to that difference in mortality.
Nature 582, 18-19 (2020)