Yes, at least five randomized controlled studies say hydroxychloroquine doesn’t help

President Donald Trump and top health officials in his administration are now openly at odds over hydroxychloroquine.

Both Anthony Fauci, the director of the National Institute of Allergy and Infectious Diseases, and Deborah Birx, coronavirus response coordinator for the White House task force, have said careful studies show the drug doesn’t speed recovery from COVID-19, the disease caused by the coronavirus.

Still, at an Aug. 3 White House event, Trump said he has taken hydroxychloroquine and it is "very highly thought of."  Trump said the drug is politically toxic because he supports it.

On NBC’s Meet the Press, Adm. Brett Giroir, the assistant secretary of health and overseer of the virus testing effort, said it is time to "move on."

"There have been five randomized controlled, placebo controlled trials that do not show any benefit to Hydroxychloroquine," Giroir said Aug. 2. "So, at this point in time, we don't recommend that as a treatment. "

We asked Giroir’s office to name the five studies he referenced in that interview. Not all of them used placebos, but they all randomly assigned patients to hydroxychloroquine treatment and conventional treatment groups. The drug proved itself in none of them.

Before we summarize the studies, a brief word about why this sort of study matters.

If you want to know if a drug works, ideally the only difference between a patient who gets it and one who doesn’t is — no surprise — the drug itself. That’s harder to do than meets the eye. People aren’t identical. You can weigh them, look at their medical histories, check their blood oxygen levels and temperature, and yet, you still might miss something.

That’s where randomly assigning patients to treatment and non-treatment groups comes in.

"You don’t know all the possible things going on in the background and all that noise is what  randomizing helps filter out," said Boston University infectious disease modeler Brooke Nichols. "It’s the best way to be sure that the only effect is the effect of the drug."

Letting a computer program assign patients also prevents any subconscious choices by doctors from distorting the results. The whole approach is designed to take human foibles out of the picture.

To really minimize the human factor, not only do you randomly assign patients to groups, but the patients and the health care workers themselves don’t know what group they are in. So everyone gets a pill, but for half of them, the pill might be just sugar or something equally neutral. That neutral pill is a placebo.

"That’s the gold-gold standard of research," said Nichols. "And just one one small notch below that is the randomized controlled trial without a placebo."

Nichols said that the lack of a placebo takes on more importance when a non-placebo study comes up with different results than one with a placebo.

"When the results are in agreement, it doesn’t matter as much," Nichols said.

Giroir’s office sent us a list of the studies he had in mind. All of them used the randomized control approach. Two of them used placebos, and three did not. They all found the drug ineffective. Here’s Giroir’s list (plus a couple more we found ourselves):

A Cluster-Randomized Trial of Hydroxychloroquine as Prevention of Covid-19 Transmission and Disease, MedRxIV, July 26, 2020. No placebo. This study out of Spain aimed to see if the drug prevented the disease. It involved about 2,300 people who had been exposed to someone who tested positive for the virus. After random assignment, one group got a dose of hydroxychloroquine for a week, and the other didn’t.

Featured Fact-check

Donald Trump

stated on October 23, 2024 in a post on X

Vice President Kamala Harris has been “going so far as to call me Adolf Hitler.”

The differences were small, too small to be statistically significant. In the treatment group, 5.7% caught the virus. In the non-treatment group, 6.2% did. 

"Post-exposure therapy with hydroxychloroquine did not prevent SARS-CoV-2 disease and infection," the authors wrote.

Hydroxychloroquine with or without Azithromycin in Mild-to-Moderate Covid-19, New England Journal of Medicine, July 23, 2020. No placebo. This study from Brazil assigned 665 patients (504 with testing-confirmed COVID-19) to three groups: standard care, hydroxychloroquine plus azithromycin and hydroxychloroquine alone. 

"Among patients hospitalized with mild-to-moderate COVID-19, the use of hydroxychloroquine, alone or with azithromycin, did not improve clinical status at 15 days as compared with standard care," the authors wrote.

Hydroxychloroquine in Nonhospitalized Adults With Early COVID-19, Annals of Internal Medicine, July 16, 2020. Placebo used. This study of 491 patients in the United States and Canada used two groups: one receiving hydroxychloroquine and the other a placebo. Participants either tested positive, or had the symptoms of COVID-19 and lived or worked with someone who had tested positive. They could not have had symptoms for more than four days.

"We hypothesized that starting hydroxychloroquine therapy within the first few days of symptoms could alter the course of COVID-19 by reducing symptom severity and duration and preventing hospitalizations," the authors wrote.

They found that the severity of symptoms after 14 days "did not differ between the hydroxychloroquine and placebo groups." 

No clinical benefit from use of hydroxychloroquine in hospitalised patients with COVID-19, Recovery Trial - University of Oxford, June 5, 2020. No placebo. This study out of the United Kingdom randomly assigned over 4,600 patients hospitalized with COVID-19 to a hydroxychloroquine treatment group and a standard care group. Researchers measured the death rate and symptoms after 28 days.

Among the treatment group, 25.7% died, compared to 23.5% for the standard care group. That was not statistically different. And there were no "beneficial effects on hospital stay duration or other outcomes."

"Although it is disappointing that this treatment has been shown to be ineffective, it does allow us to focus care and research on more promising drugs.," the researchers wrote.

The data has yet to be released.

A Randomized Trial of Hydroxychloroquine as Postexposure Prophylaxis for Covid-19, New England Journal of Medicine, June 3, 2020. Placebo used. This study focused on hydroxychloroquine as a way to avoid getting the disease. They randomly assigned 821 people with no symptoms — but who had spent time close to a person who tested positive — to a treatment or placebo group.

The hydroxychloroquine group was slightly less likely to test positive after two weeks, 11.8% compared to 14.3% in the placebo group, but the numbers were too small to pass statistical muster.

We found two other randomized controlled studies in addition to those Grigoir referenced. One out of Spain worked with 293 people with confirmed infections, but who were not hospitalized. There was no placebo. They found that hydroxychloroquine did not cut the risk of hospitalization, or help people recover more quickly.

An early study in May in Shanghai, China, assigned 75 COVID-19 patients to a hydroxychloroquine treatment group, and another received typical care. It also found no difference in the speed of recovery.

There has been one large study in Detroit that concluded that the drug worked. However, patient data in that study was collected after the fact and there was no randomized control process.

Giroir said there were five "randomized controlled, placebo controlled trials that do not show any benefit to hydroxychloroquine."

He is largely correct. There have been more than five randomized controlled trials that found hydroxychloroquine ineffective, either at preventing people from catching the virus, or helping them recover if they become ill. He exaggerated the number of studies that used placebos — out of his list, two did and three did not.

The infectious disease modeler we reached said placebos improve the quality of a study, but the lack of one matters less if the results are in line with placebo-driven trials.

We rate this claim Mostly True.