Gilead’s $1,000-a-pill antiviral remdesivir is no wonder drug. We knew this when it failed for hepatitis, the disease it was created for. And then when it failed for Ebola. And then again, when it failed for COVID-19.

But like a bolt, in late April, a breakthrough: in a second trial for COVID-19 patients, remdesivir sped time to recovery—proof of benefit, according to Anthony Fauci. Since his press briefing, which offered a narrow glimpse of the study’s results, scientists and doctors, many of them skeptics based on the drug’s litany of failures, have been eagerly awaiting a peer-reviewed report.

Meanwhile, some have lambasted the delay, while others have pointed to irregularities in the study registry, including a change in its primary outcome—a major red flag. It turns out the researchers planned and designed their trial to achieve one goal, then moved the goalposts midway through the study. This is broadly considered inappropriate and has the potential to invalidate any findings in the minds of influential voices in the medical community. But in rare cases, such a maneuver can be defensible. And the only way to know is to hear precisely how and why it happened. More than anything, this is what hawkish researchers and skeptics were looking for from a full study report.

They didn’t get it.

On May 22, the New England Journal of Medicine published “Remdesivir for the treatment of COVID-19—Preliminary Report.” Even the title seems like a disclaimer, suggesting a fuller account will be forthcoming. Well, it better be.

For starters, incredibly, the paper never even mentions how patients were selected for the trial. No age range, no severity criteria, no care settings (were patients enrolled from nursing homes? hospitals? clinics?), no mention of who was eligible, or ineligible. Nothing. This alone makes it impossible for doctors to know how they might apply the study’s results in practice.

The bigger gap, however, is a milquetoast paragraph saying close to nothing about their explosively controversial midstream change. Originally their primary outcome was a scale assessing deaths, ventilator needs, and other outcomes from COVID-19. When they learned the disease had a “more protracted course,” they write, they switched their primary outcome to “time to recovery.” They were concerned, they say, “a difference … after day 15 would have been missed by a single assessment at day 15.”

Tautologies aside, this brings us no closer to understanding their motives. Had they simply desired more time to capture outcomes from COVID-19 as they claim, they need not have changed their outcome. They could simply have made their measurement two weeks later, on day 29 instead of day 15. The 29-day assessment was already in their protocol and would have captured the overwhelming majority of important outcomes. It would have taken no extra effort, and meant no outcome change, no new power calculation, no statistical gymnastics, and no new endpoints or calculations. Shifting the date of assessment by two weeks could have solved the problem.

Moreover, this would have respected the fact that patients and health authorities everywhere are far more interested in saving lives and avoiding the need for ventilation. The new outcome choice, time to recovery, is bizarrely aloof to hundreds of thousands of deaths, and rampant fears about ventilator shortages. This is a deadly pandemic. No one is clicking their heels and praying for a drug that improves “time-to-recovery.”

They also fail to explain why, when they made the change, they didn’t use one of the time-to-recovery measures they had planned all along as secondary outcomes. Instead, mid-trial, they created whole a new definition and calculation for time to recovery, raising the question of what they knew about their own data and why the measures they devised and trusted before the trial were suddenly inadequate.

Thus even after reading the paper, doctors around the world cannot glean whether remdesivir works. With a history of nothing but failure and now, one nominally successful study chock full of mysterious irregularities, there is even more to doubt.

If the NIAID researchers believe in the utility of this pricey drug, the medical community will need more than a press release, and more than an obfuscating preliminary report, to be convinced this trial isn’t just another remdesivir failure.

Daniel Hopkins is a physician.

Image credit: Shutterstock.com