President Donald Trump has urged doctors and patients to treat coronavirus infections with hydroxychloroquine, a drug that is not approved for that purpose.

“What do you have to lose?” he asked Saturday.

The answer to that question is not straightforward.

Set aside the facts that the drug is needed to treat other diseases, such as lupus, for which it has been proven to help patients, and that the president does not have medical expertise. Physicians nevertheless have been using hydroxychloroquine to treat certain coronavirus patients with severe symptoms, and some have gotten better.

Using a drug to treat a disease for which it is not approved is called using it off-label. The practice is perfectly legal, and it is understandable why families would push for such a drug if no alternatives were available.

Here’s the downside. When a patient takes a drug, one of three things can happen:

  • It will enhance the person’s recovery.

  • It will interfere with recovery.

  • It will have no impact.

If every patient were to “try it if you like,” as Trump urged, we would never know which of these three things had happened. Just because a person gets better after taking a drug, it does not mean the drug helped; it’s possible the patient would have gotten better faster without it.

For society at large, scientists say, the better approach is what’s called a clinical trial.

» FAQ: Your coronavirus questions, answered.

Before approving a drug for a particular use, the U.S. Food and Drug Administration requires evidence from such trials: studies in which some patients are given the drug in question, while others are given some other drug or a placebo, depending on the circumstances.

Patients are followed for a predetermined period to see if one group is the more likely to show signs of improvement.

» READ MORE: Experimental coronavirus vaccine given to Philadelphia volunteers

Ideally, patients are randomly assigned to these categories, to make sure that the impact of the drug can be measured. If the experimental drug was given only to those who asked for it, that could skew the results. (For example, those patients might share some characteristic that would help them recover — say, higher socioeconomic status — but not be known to the scientists. If such patients got better after taking the drug, it could be due to the fact that they have more resources.)

Add the fact that drugs, whether or not they have any impact on the disease in question, can also cause harmful side effects, and there is plenty of reason for caution.

With hydroxychloroquine, those ill effects can include a dangerous irregular heart rhythm. That side effect does not occur in most people who take the drug, but may be an unnecessary risk when balanced against the drug’s unclear benefit.

Hydroxychloroquine is a cousin of chloroquine, a drug that was developed to treat malaria in the 1940s. The reason some physicians think it could work against the coronavirus is because it raises the pH inside human cells, potentially making it harder for the virus to replicate.

But this month, one study that purported to show the drug’s benefit drew fire from the International Society of Antimicrobial Chemotherapy — the professional organization that oversees the journal in which it was published.

“The ISAC board believes the article does not meet the society’s expected standard,” the group said.

» READ MORE: Coronavirus debate: Could blood pressure meds make COVID-19 less — or more — deadly?

Among other issues, the article gives a poor explanation of why certain patients were excluded, the board wrote. In academia, that’s a potential red flag. If scientists cherry-pick which participants are included in their final results, a drug can seem to have an impact when it does not.

Another study of hydroxychloroquine in coronavirus patients was prematurely halted with little explanation — another possible red flag, said Susan Ellenberg, a professor of biostatistics at the University of Pennsylvania’s Perelman School of Medicine. The drug appeared to have a modest benefit at the moment the trial was cut short, but there were not enough data to indicate whether that improvement was simply due to chance, she said.

“If you’re doing a clinical trial, the data oscillate," she said. “You need to have enough data to be pretty sure you’re not just seeing a random high.”

Among the coronavirus patients who appreciate the distinction is Maureen Boland, the Abington High School teacher who was treated with hydroxychloroquine and wrote about her experience.

After Boland was admitted to a hospital, which she did not identify, her family debated with doctors about whether to try the drug, she wrote.

“Whether it was the hydroxychloroquine or not, one day after being given the drug combination (antibiotics were added), I made some progress and began breathing better,” Boland said.

» READ MORE: Why the coronavirus and most other viruses have no cure

But was it the drug that worked? Or was it the nebulizer and other treatments she was given?

Until formal, rigorous studies are performed, we will not know.