Read our COVID-19 research and news.

Participants in a hydroxychloroquine study enrolled on a website and received medicines overnight by FedEx.

John Phillips/Getty Images

‘I would not recommend this.’ A scientist’s hydroxychloroquine trial—and his advice to Trump

Sciences COVID-19 reporting is supported by the Pulitzer Center.

Yesterday, The New England Journal of Medicine (NEJM) published the highly anticipated results of a clinical trial of hydroxychloroquine to prevent COVID-19. The study, led by David Boulware of the University of Minnesota, Twin Cities, enrolled 821 healthy adults who were at risk of the disease: They had been in close proximity to a COVID-19 patient for more than 10 minutes, either at home or at work, without wearing full protection.

The trial showed no benefit from hydroxychloroquine; 12% of the people who took the drug went on to develop symptoms of COVID-19, versus 14% in a placebo group, a difference that was not statistically significant. But because of the small size and other features of the trial, an editorial published alongside the study called the results “more provocative than definitive.” 

Science talked to Boulware about the limitations of the “internet-based” study, the future of research on hydroxychloroquine—complicated today by the retraction of a major study that suggested the drug is dangerous—and the advice he gave when President Donald Trump’s physician called him. Questions and answers have been edited for brevity and clarity.

Q: Why did you decide to do a prevention trial?

A: In early March, as community spread was just beginning in the U.S., no one was doing prevention studies. Once people are in the hospital you may be able to keep them from dying, but if you think of how you’re going to change the pandemic, breaking the chain of infection is how you’re going to [do that].

Q: And why did you choose to use hydroxychloroquine?

A: I was aware that it has activity against SARS-CoV-2 in cell culture. That does doesn’t necessarily mean it’s going to work in humans and so we wanted to do a trial. Hydroxychloroquine was available. It’s cheap. It’s safe. Most of my work is international and this is internationally available.

Q: You didn’t have a laboratory diagnosis for all of the people who went on to develop COVID-19 symptoms, right?

A: In March and April, there were no tests available for outpatients in the U.S. And so even health care workers, who should have been tested by national guidelines, could not get tests. It would have been great if the U.S. had testing. So that’s a limitation.

Q: It means that you can’t be 100% certain that what you’re comparing is the incidence of COVID-19 in both groups?

A: That’s why you have a randomized trial. Could there be people with influenza or rhinovirus? Yeah, there certainly could be. But they would be equally distributed across both arms, so that should wash out. And we’re not just talking about random people off the street, but people with a high-risk epidemiological exposure.

Q: This was not a typical trial in how the drugs were administered.

A: It was an internet-based trial. So people could enroll anywhere in North America on a website, they got screened, and if they were eligible we FedExed the medicines overnight. And then they took the medicine and we did follow-up surveys.

There’s a lot of media saying hydroxychloroquine is dangerous, so 15% of people in both arms didn’t even start the medicine because they were freaked out by the side effects. And of the people who started it, there was a higher dropout rate in the hydroxychloroquine arm. But even when we looked at the group that just took the medicine, there still was not any benefit.

Q: One of the big discussion points, of course, has been the side effects of hydroxychloroquine.

A: Forty percent of people in the hydroxychloroquine arm had some side effects, but nothing serious. There was no hospitalization, no cardiac arrhythmias. There were side effects as expected—gastrointestinal symptoms, nausea, upset stomach—and most of those were mild and went away.

The most interesting thing was that 17% of the people in the placebo group reported side effects. They were describing vertigo and tinnitus and all the things that were on the package insert. I found that really fascinating. Our placebo was folic acid, so it shouldn’t do any of those things.

Q: Generally, the earlier you give a drug against a virus, the more likely you are to see an effect. So if there is no effect in a postexposure prophylaxis study, doesn’t that make it much less likely that there will be an effect in really sick patients?

A: Well, there may be anti-inflammatory things that hydroxychloroquine does, or there may be other reasons why it might work. But yeah, it does put a big kind of red flag on whether this is going to work. Even if you look at remdesivir, [considered the most promising COVID-19 treatment so far], it didn’t really do anything in really sick patients. We need more trials and properly powered trials and randomized trials, double-blind trials, that can help inform stuff.

Q: We’re getting a lot of these smaller trials. Aren’t we better off pooling all of these resources and doing really big trials? The NEJM editorial by Myron Cohen called the results “more provocative than definitive.”

A: I was sort of surprised by what he wrote at first, but the larger point he is making is that we need more large randomized clinical trials. But our trial was a $100,000 trial, run on a shoestring budget. We didn’t wait for funding, we just did it.

Q: We all look to science ideally to give us clear answers. But I don’t see how anyone on either side of this will take this trial and say: “This convinces me of the other point of view.”

A: We would say for postexposure prophylaxis, there’s no evidence that hydroxychloroquine can prevent disease. So if you’ve been exposed to someone, I would say a physician should not be prescribing this. There’s no data to support it. … Hospitalized patients, that’s a separate issue.

Q: So what other studies do you think will give us answers?

A: There’s a trial in Barcelona[, Spain,] for postexposure prophylaxis that should finish in a few weeks. We’ll see if they show something similar as we do. WHO’s [the World Health Organization’s] Solidarity trial, which is a very good trial, will have data for treatment. Hydroxychloroquine has become such a political topic that in the U.S. you can’t get people to enroll in trials with it. So most of the U.S. trials are struggling.

Q: Why?

A: For the people who believe that works: Why would you be in a placebo-controlled trial? And for people who believe it doesn’t work: Why would you be in a trial with hydroxychloroquine?

Q: Can we agree that Trump shouldn’t be taking hydroxychloroquine?

A: Correct. I was asked by Trump’s doctor about it and I recommended no.

Q: When did he ask you?

A: It was May 9, the day we submitted the manuscript to [NEJM]. It was right when he was exposed to his valet, so they were reaching out for data. I was asked by the White House medical unit director what our dose was. I shared the dose we’re using and our safety data. And I said, based on the public data, “I would not recommend this.” But I also said: “I realize that you’re going to ignore my opinion, because Trump has been talking about this for 2 months.”

*Correction, 6 June, 3:55 p.m.: In a previous version of this story, Boulware said Myron Cohen is an investigator on a $9 million trial funded by the U.S. government. This was incorrect; that answer has been changed.