The first large-scale study showed that hydroxychloroquine was no better than placebo pills to prevent coronavirus disease


A malaria drug President Donald Trump took to try to prevent COVID-19 proved ineffective for that in the first large, high-quality study to test it in people in close contact with someone with the disease.

Results published Wednesday by the New England Journal of Medicine show that hydroxychloroquine was no better than placebo pills at preventing illness from the coronavirus.

The drug did not seem to cause serious harm, though—about 40% on it had side effects, mostly mild stomach problems.

“We were disappointed. We would have liked for this to work,” said the study leader, Dr. David Boulware, an infectious disease specialist at the University of Minnesota.

“But our objective was to answer the question and to conduct a high-quality study,” because the evidence on the drug so far has been inconclusive, he said.

Hydroxychloroquine and a similar drug, chloroquine, have been the subject of much debate since Trump started promoting them in March. Hydroxychloroquine has long been used for malaria, lupus and rheumatoid arthritis, but no large studies have shown it or chloroquine to be safe or effective for much sicker patients with coronavirus, and some studies have suggested the drugs may do harm.

Trump took a two-week course of hydroxychloroquine, along with zinc and Vitamin D, after two staffers tested positive for COVID-19, and had no ill effects, according to results of his latest physical released by his doctor Wednesday.

Federal regulators have warned against hydroxychloroquine’s use except in hospitals and formal studies because of the risk of side effects, especially heart rhythm problems.

Boulware’s study involved 821 people in the United States and Canada living with someone diagnosed with COVID-19 or at high risk of getting it because of their job – doctors, nurses, ambulance workers who had significant exposure to a sick patient while not wearing full protective gear.

They were randomly assigned to get either the nutrient folate as a placebo or hydroxychloroquine for five days, starting within four days of their exposure. Neither they nor others involved in the research knew who was getting which pills.

After 14 days in the study, 12% on the drug developed COVID-19 symptoms versus 14% in the placebo group, but the difference is so small it could have occurred by chance, Boulware said.

“There’s basically no effect. It does not prevent infection,” he said of the drug. Even if it were to give some slim advantage, “we’d want a much larger effect” to justify its use and risk of side effects for preventing illness, he said.

Results were no different among a subgroup of participants who were taking zinc or vitamin C, which some people believe might help make hydroxychloroquine more effective or fight the coronavirus.

There are some big caveats: The study enrolled people through the Internet and social media, relying on them to report their own symptoms rather than having them tracked in a formal way by doctors.

Participants were not all tested for the coronavirus but were diagnosed as COVID-19 cases based on symptoms in many cases. And not all took their medicines as directed.

The results “are more provocative than definitive,” and the drug may yet have prevention benefits if tried sooner or in a different way, Dr. Myron Cohen of the University of North Carolina at Chapel Hill wrote in a commentary in the journal.

Others were glad to see a study that had a comparison group and good scientific methods after so many weaker reports on hydroxychloroquine.

“This fits with everything else we’ve seen so far which suggests that it’s not beneficial,” said Dr. Peter Bach, director of a health policy center at Memorial Sloan Kettering Cancer Center in New York.

This study was in younger relatively healthy people, but the results “would make me very discouraged about trying to use this in older people” who are most vulnerable to serious illness from the coronavirus, Bach said. “If it does work, it doesn’t work very well.”

Dr. Dan Culver, a lung specialist at the Cleveland Clinic, said there’s still a chance that giving the drug sooner than four days after someone’s exposure to the virus may help prevent illness.

But the study “takes ‘home run’ off the table” as far as hopes for the drug, he said.

The study was mostly funded by David Baszucki, founder of Roblox, a California-based game software company, and other private donors and the Minnesota university.

Boulware also is leading a study testing hydroxychloroquine for treating COVID-19. The study is finished and results are being analyzed now.

On Tuesday, the journal Lancet posted an “expression of concern” about a study it published earlier this month of nearly 15,000 COVID-19 patients on the malaria drugs that tied their use to a higher risk of dying in the hospital or developing a heartbeat problem.

Scientists have raised serious questions about the database used for that study, and its authors have launched an independent audit.

That work had a big impact: the World Health Organization suspended use of hydroxychloroquine in a study it is leading, and French officials stopped the drug’s use in hospitals. On Wednesday, the WHO said experts who reviewed safety information decided that its study could resume.



We recruited 821 asymptomatic adult participants who were randomly assigned to the hydroxychloroquine group (414 participants) or the placebo group (407 participants) (Figure 1).

The demographic and clinical characteristics of the participants are provided in Table 1. The median age was 40 years (interquartile range, 33 to 50). Women accounted for 51.6% of the trial participants (424 of 821).

A total of 27.4% of the participants (225 of 821) reported chronic health conditions, with hypertension being the most common (99 of 821 [12.1%]), followed by asthma (62 of 821 [7.6%]).

Health care workers accounted for 66.4% of the participants (545 of 821), the majority being physicians or physician assistants (342 of 545 [62.8%]) and nurses or nursing assistants (128 of 545 [23.5%]).

In the case of health care workers, exposure was predominantly from patients (418 of 545 [76.7%]) or ill coworkers (107 of 545 [19.6%]). Among the 29.8% of the participants (245 of 821) who enrolled as a household contact, the majority reported that their Covid-19 contact exposure was either a spouse or partner (114 of 245 [46.5%]) or a parent (43 of 245 [17.6%]).

Overall, 87.6% of the participants (719 of 821) had high-risk exposures without eye shields and surgical masks or respirators. Of those, 365 received hydroxychloroquine and 354 received placebo. Approximately 60% of the participants reported not wearing any element of personal protective equipment during their Covid-19 exposure.


Overall, new Covid-19 (either PCR-confirmed or symptomatically compatible) developed in 107 of 821 participants (13.0%) during the 14 days of follow-up (Table 2). The incidence of new illness compatible with Covid-19 did not differ significantly between those receiving hydroxychloroquine (49 of 414 [11.8%]) and those receiving placebo (58 of 407 [14.3%]) (P=0.35).

The absolute difference was −2.4 percentage points (95% confidence interval, −7.0 to 2.2). Figure 2 shows the development of Covid-19 over time. Two hospitalizations were reported (one in each group). No arrhythmias or deaths occurred.

There was no meaningful difference in effectiveness according to the time of starting postexposure prophylaxis or in any of the prespecified subgroups (Fig. S1 in the Supplementary Appendix).

Overall, 10.7% of the participants (46 in the hydroxychloroquine group and 42 in the placebo group) did not complete the day 14 survey; among these participants, vital status was unknown for 36 in the hydroxychloroquine group and 33 in the control group. In sensitivity analyses, exclusion of these persons from the denominator or inclusion of them as having had an event did not affect the trial conclusions (Table S1).

Of 113 persons in whom symptomatic illness developed, 16 had PCR-confirmed disease, 74 had illness that was compatible with probable Covid-19 per the U.S. case definition, 13 had possible Covid-19 with compatible symptoms and epidemiologic linkage, and 10 were adjudicated as not having Covid-19 on the basis of the symptom complex (Table S2).

Four additional participants had positive PCR tests and were asymptomatic during the 14-day trial period; symptoms eventually developed in 3 of these participants. The median number of symptoms was 4 (interquartile range, 2 to 5) among participants with Covid-19. The most frequent symptoms were cough (44.9% of the 107 participants with Covid-19), fever (34.6%), shortness of breath (18.7%), fatigue (49.5%), sore throat (40.2%), myalgia (37.4%), and anosmia (23.4%).

Among participants who were symptomatic at day 14, the median symptom-severity score (on a scale from 0 to 10, with higher scores indicating greater severity) was 2.8 (interquartile range, 1.6 to 5.0) in those receiving hydroxychloroquine and 2.7 (interquartile range, 1.4 to 4.8) in those receiving placebo (P=0.34).


Adherence among the trial participants was moderate. Full adherence to the trial intervention differed according to trial group, with 75.4% of participants in the hydroxychloroquine group (312 of 414) and 82.6% of those in the placebo group (336 of 407) having taken all 19 prescribed tablets over a period of 5 days (P=0.01).

The most common reason that participants stopped taking the assigned hydroxychloroquine or placebo was side effects (17 participants in the hydroxychloroquine group and 8 in the placebo group). Side effects were more frequent with hydroxychloroquine than with placebo (Table 3).

Among the participants who took any hydroxychloroquine, 40.1% (140 of 349) reported a side effect by day 5, as compared with 16.8% (59 of 351) receiving placebo (P<0.001). Nausea, loose stools, and abdominal discomfort were the most common side effects. There were no serious intervention-related adverse reactions or cardiac arrhythmias.

On day 14, we assessed how well the masking of the trial interventions was maintained. Of the 344 participants in the hydroxychloroquine group who completed the day 14 survey question, 160 (46.5%) correctly identified that they received hydroxychloroquine, 151 (43.9%) were unsure, and 33 (10%) believed that they received placebo.

Of the 353 participants in the control group who completed the day 14 survey question, 126 (35.7%) correctly identified that they received placebo, 168 (47.6%) were unsure, and 59 (16.7%) believed that they received hydroxychloroquine.

Participants who reported any side effect (regardless of trial group) at day 5 were 3.7 times as likely to believe that they received hydroxychloroquine as participants who did not report side effects (122 of 179 participants [68.2%] reporting side effects and 94 of 504 participants [18.7%] not reporting side effects; P<0.001). In the absence of side effects, blinding was well maintained.

Supplementary Material

Supplementary AppendixPDF270KB
Disclosure FormsPDF405KB
Data Sharing StatementPDF69KB

Journal information:New England Journal of Medicine, The Lancet


Please enter your comment!
Please enter your name here

Questo sito usa Akismet per ridurre lo spam. Scopri come i tuoi dati vengono elaborati.