A spoonful of coronavirus medicine

Several readers have wondered about proposed treatments for the coronavirus. One asked this question: How did the experimental drug remdesivir suddenly supplant hydroxychloroquine as the most likely candidate to help COVID-19 patients? If remdesivir helps and hydroxychloroquine doesn’t, why haven’t all the studies shown that?

The short answer first: We’re dealing with preliminary studies, each with shortcomings. All studies have a hypothesis (an idea to test). Good medical studies generally have a control group (a group of participants who don’t get the proposed treatment), assign patients randomly to a treatment or placebo, and analyze all the relevant data the experiment produces. (Good studies share several other characteristics, too.)

With those qualities in mind, let’s look at the relevant studies examining whether hydroxychloroquine and remdesivir are beneficial for people suffering from COVID-19.

One small French study appeared to support the use of hydroxychloroquine for COVID-19, but the study had no control group. Worse, the authors excluded patients who worsened while on the drug. That wasn’t good research technique: If we want to know whether a drug helps people recover, drug recipients who instead get sicker are actually a valuable clue.

A larger U.S. study of Veterans Health Administration patients who had the coronavirus appeared to show that hydroxychloroquine didn’t help and might even cause harm. But the VA was only giving hydroxychloroquine to the sickest patients: As with any disease, sicker patients don’t do as well on average. The patients also received an “antiviral” dose of the drug, a much higher dose than patients with other conditions would get. That can lead to problems—especially in older, less healthy patients like those in the VA study. The VA study referenced Brazilian research that found the closely related drug chloroquine increased the likelihood of death when given in high doses, most likely due to effects on the heart. (Kudos to the researchers for exploring what their data might signify, even if it wasn’t what they’d hoped for.)

The research team in France responded with a larger retrospective study, but it has two major flaws. The first flaw is recruiting young, otherwise healthy patients: The COVID-19 coronavirus prefers to attack the elderly, so good outcomes in patients averaging 43 years of age may not be noteworthy. (By comparison, the VA study’s average age was 69—and its patients were in poorer overall health to begin with.) The second flaw prevents us from knowing whether those young patients treated with hydroxychloroquine would have recovered anyway: There’s still no control group! Without a control group, we have no way of knowing whether the treatment helped, harmed, or didn’t do anything at all.

How about remdesivir, then? A well-designed study in China planned to compare the drug with a placebo, but researchers suspended it after failing to enroll as many patients as they’d intended. Preliminary data posted to the World Health Organization’s website, then later deleted, did not appear to show a benefit to patients, dampening enthusiasm. Drugmaker Gilead has its own large study, but it lacks a control group.

More recently, the Adaptive COVID-19 Treatment Trial, sponsored by the National Institutes of Health (NIH), released preliminary results appearing to support the use of remdesivir, but with a catch. The study’s original goal was to demonstrate reduced mortality: The drug appeared to do so, but never “reached statistical significance.” In other words, the study’s outcomes could have been due simply to chance. The study’s new goal assessed the length of patient hospitalization, and remdesivir appeared to shorten that—but moving the goalposts midgame suggests researchers may have been unimpressed with their early results.

Ending with a plea for more research is a common trope in writing about research, but it’s more relevant now than ever. The currently published research is the preliminary work that happens while we’re racing to determine what’s worth further study. Based on what I’ve seen, I’d focus on the VA study for hydroxychloroquine and the NIH study for remdesivir—their control groups make them more reliable. My conclusion: Remdesivir appears to have a modest, if somewhat weak, benefit for COVID-19 patients. Also, high-dose hydroxychloroquine appears to carry more health risks than the low-dose version given to lupus patients.

In the meantime, wear your mask and wash your hands. An ounce of coronavirus prevention may still be worth a pound of cure.