A COVID drug helped patients recover 5 days sooner but barely cut deaths

In the first year of the COVID-19 pandemic, one drug went from headline hope to headline doubt within months. Two of the biggest trials of that time had tested it in the same kind of patient, adults sick enough to be in hospital [1][2].

In the first trial, of 1,062 patients, the typical patient on the drug was ready to leave hospital in 10 days. For patients given a dummy drip with no medicine in it, it took 15 days. That is five days sooner [1].

The catch is that getting better faster is not the same as fewer deaths. A much bigger trial, of 8,275 people, found a drop in deaths too small to be sure of overall. The two trials were asking different questions [3].

The drug is remdesivir, made by Gilead Sciences. A virus spreads by copying its genes. Remdesivir looks like one of the small pieces the virus uses to build those copies, so when the virus grabs it, the copying gets stuck. The United States allowed doctors to use it in an emergency in May 2020 [4].

The first trial, called ACTT-1, was led by John Beigel of the US National Institute of Allergy and Infectious Diseases, which also paid for it. It was published in the New England Journal of Medicine in 2020. Chance decided, like a coin flip, who got the drug. Neither the patients nor their doctors knew who got the real drug and who got the dummy drip [1].

The researchers put their main finding this way:

Our data show that remdesivir was superior to placebo in shortening the time to recovery in adults who were hospitalized with Covid-19 and had evidence of lower respiratory tract infection.

In plain words, patients with an infection deep in the lungs got home sooner on the drug than on the dummy drip. By day 29, deaths were lower too, about 11 in 100 against about 15 in 100. But a trial this size was too small to say that gap in deaths was real [1].

So why did the bigger trial look different?

The second trial, called Solidarity, was run and paid for by the World Health Organization. It signed up patients at 454 hospitals in 35 countries and asked one simple question: do fewer people die? There was no dummy drip, so everyone knew who got the drug [3].

Its first report came out in the New England Journal of Medicine in 2021. It found that 301 of 2,743 people died on the drug, and 303 of 2,708 died without it. That is almost no difference [2].

The final report, in The Lancet in 2022, covered 8,275 people. Overall, 14.5 percent died on the drug, against 15.6 percent without it, a gap too small to rule out chance. Among people not yet on a breathing machine, it was 11.9 percent against 13.5 percent, a gap the trial counted as real. For people already on a machine, the drug did not help [3].

The team summed it up this way:

Among other hospitalised patients, it has a small effect against death or progression to ventilation (or both).

In plain words, among patients not yet on a breathing machine, slightly fewer died or ended up needing one. For those already on one, the drug made no clear difference [3].

How can both trials be right? A trial of about a thousand people can spot a five day gap in getting better. But it is far too small to settle a question about deaths. A trial of eight thousand can settle deaths, but it never tracked how fast people got better [1][3].

Who knew what also mattered. In an open trial, knowing who got the drug could nudge when a doctor sends someone home. That is one reason Solidarity counted deaths, the result hardest to nudge by knowing [2][3].

Care also changed between the two trials. By the time Solidarity ended, doctors were giving patients who needed oxygen a cheap steroid, a medicine that calms swelling in the body. When usual care is already better, an added drug has less room to show a difference [3].

These trials do not show what the drug does against the kinds of the virus going around now, when many people have some protection from earlier shots or infections. They also do not cover people treated at home. Every number here comes from patients sick enough to be in hospital in 2020 and 2021 [1][2][3].

What the trials do show is smaller than the early headlines and bigger than the later ones. One careful trial found faster recovery. A much larger one found a small drop in deaths for hospital patients not yet on a breathing machine, and none for those already on one. Neither trial was wrong. Each was built to answer a different question [1][3].