About 88 percent had high-risk exposures.
The participants, recruited online, ranged in age from 33 to 50, with a median age of 40. About half were women, and 66 percent of the total were health care workers. They were healthy and had no underlying health problems that would have made hydroxychloroquine dangerous for them. Most of the rest had been exposed at home, to an infected spouse, partner or parent.
Within four days of exposure, the participants were picked at random to receive either hydroxychloroquine or a placebo, and then followed to determine whether they had either laboratory-confirmed Covid-19, or an illness consistent with the virus, during the next 14 days.
The drug or placebos were mailed to them, and they then reported their symptoms online to the researchers, who did not examine them.
Not all the participants could be tested for the virus, because when the study was being conducted, there was still a shortage of test kits.
There was no meaningful difference between the placebo group and those who took the drug. Among those taking hydroxychloroquine, 49 of 414, or 11.8 percent, became ill. In the placebo group, 58 or 407, or 14.3 percent, became ill. Analyzed statistically, the difference between those rates was not significant.
The drug also did not make the illness any less severe.
Side effects like nausea from hydroxychloroquine were more common than from placebos, 40.1 percent compared with 16.8 percent, but there were no problems with heart rhythm or any other serious adverse effects.
Infectious disease experts who were not part of the study said it was well done and answered an important question, though the results were disappointing.