(Bloomberg) -- Almost three months after President Donald Trump first touted an old malaria medicine as a wonder drug that could change the course of the pandemic, conclusive evidence that it works -- or not -- appears equally shaky.

The confusion reached new heights in recent days. On Wednesday, scientists said the treatment, hydroxychloroquine, didn’t work to prevent Covid-19 -- which doesn’t speak to its potential as a treatment. But the next day, a large study from May that linked the medication to cardiovascular risks was retracted after it turned out to rely on questionable data.

Clashes over the aging drug’s merits have emerged as one bizarre twist of this pandemic, with some countries stockpiling hydroxychloroquine and others warning of lethal side effects. As politics and science collide, doctors are left to make treatment decisions without sound, impartial data.

“It makes no sense in the middle of a pandemic that we aren’t laser-focused on learning as quickly as possible,” said Peter Bach, director of the Center for Health Policy and Outcomes at Memorial Sloan Kettering Cancer Center. “It’s frustrating. It matters. There are people dying.”

When the Lancet published a study of about 96,000 patient records on May 22 showing that the controversial medicine increased the risk of death and heart ailments, it appeared to settle the debate: If the product caused harm, it was unlikely to be a panacea for the virus. The World Health Organization swiftly stopped recruiting new people in the hydroxychloroquine portion of a broad Covid-19 clinical trial designed to size up potential cures.

However, doctors found inconsistencies in the data that called into question the study’s reliability and eventually raised doubts about other work from the same data supplier, Surgisphere Corp. The researchers, saying they had worked “in good faith and at a time of great need,” retracted the study in the Lancet late Thursday.

“There seems to be a problem with some of the world’s best medical journals,” said Philippe Parola, head of infectious diseases at IHU-Mediterrannee Infection hospital in Marseille, France. “There are many scientific papers of poor quality, but you have to wonder why this one ended up in the Lancet.”

Parola, one of the hundreds of scientists who signed a letter to the journal asking for greater transparency regarding the study data, works with Didier Raoult, the IHU’s boss and a folk hero to some in France for his iconoclastic positions and early championing of hydroxychloroquine.

Unclear Evidence

Upon seeing the alarming research published in the Lancet, at least three other groups running clinical trials on the medicine paused to examine their own data, since a greater mortality risk would warrant immediately halting research.

The WHO, Swiss drugmaker Novartis AG and the U.K.’s Wellcome Trust all said they found no such signals in their studies, suggesting that regardless of how well the drug works, the evidence doesn’t point to more deaths. A review of published studies by Cochrane, a scientific group that assesses what’s known on a given topic, also reached a similar conclusion, though it warned about the very low certainty of its evidence.

Now scientists need to go back to investigating whether there is a medical benefit to taking the treatment, a generic manufactured by several drugmakers including Novartis, Sanofi and Mylan NV. The new coronavirus has killed more than 388,000 people worldwide and there’s no cure yet.

Hydroxychloroquine shot to fame in March, propelled in part by Raoult’s work and notably one study that was criticized for its small sample size, the exclusion of some patients and the lack of a control group. Other studies bolstered the claims, finding the drug made it more difficult for the virus to infect healthy cells and showing the medicine had anti-inflammatory properties that might calm the over-active immune systems of severely ill patients.

Marseille Trial

But lately, studies have started appearing that showed it was far less effective than many doctors had hoped.

The Marseille team now has gathered data on 3,737 patients. Most people were treated early with the medicine and the antibiotic azithromycin, and the researchers said the combination reduced the chances a patient would have to be placed in intensive care, as well as the risk of dying. The paper was submitted to the Lancet and rejected, according to Parola. The journal said it doesn’t comment on articles it doesn’t publish.

The only other drug cleared for the new coronavirus, Gilead Sciences Inc.’s remdesivir, also offers rather modest results, though its safety profile appears better.

Unlike remdesivir, an experimental antiviral developed for Ebola, hydroxychloroquine has years of credentials against inflammatory illnesses such as rheumatoid arthritis and lupus. It has one rare but dire side effect: it can prompt some patients’ heart rhythm to become erratic. The risk was documented long before the virus appeared in China at the end of last year, requiring doctors to carefully select patients, but concerns have grown as the product become more widely prescribed for Covid-19.

“It’s critical to generate robust and interpretable clinical evidence in order to guide future decisions,” Peter Zuest, a spokesman for Novartis, said by email. The company will press ahead with its trial “to obtain these highly needed scientific data.”

Randomized Trial

The Novartis trial includes about 440 patients hospitalized in the U.S. Some of them are prescribed hydroxychloroquine, some take it in combination with azithromycin, and others get a placebo. The patients are assigned to the different treatment groups at random, and neither they nor their doctors know who’s getting what.

Such studies are the gold standard of clinical research and crucial to generate reliable answers, said Neil Schluger, who heads the division of pulmonary, allergy and critical-care medicine at Columbia University in New York.

Schluger and colleagues called for such unbiased research in an article in the New England Journal of Medicine last month after publishing an observational study that found the drug neither increased nor decreased the risk of death or the need for intubation. That’s when doctors have to insert a tube in the windpipe of someone who can’t breathe on her own.

“It’s easy in retrospect just to say, why didn’t we just do randomized clinical trials,” he said. “But as a clinician, I wouldn’t disparage anybody’s motives for grasping at therapies based on the thinnest of evidence” in the midst of a crisis.

Politicized Drug

Trump himself, after more than a dozen tweets or retweets endorsing hydroxychloroquine, said he was taking it as a precaution after two West Wing staff members were diagnosed with Covid-19.

Endorsements from Trump and Brazil’s President Jair Bolsonaro, both populist leaders with no background in science, may have boosted the medicine in some circles but stripped it of credibility in others, mostly their critics.

“Trump didn’t do me any favors by liking hydroxychloroquine,” France’s Raoult said in an interview on BFM TV on Thursday. “Politicization is the worst thing that can happen to a medicine.”

That’s because beliefs end up warping the science. At this point, it’s possible even a persuasive clinical trial won’t be enough to settle the debate over this medicine, according to Derek Lowe, a drug-discovery scientist and the founder of the widely-read “In the Pipeline” blog about the pharmaceutical industry.

“I think if the drug were as amazing as some of its proponents claim, we’d have seen a much stronger clinical signal by now,” Lowe said. But “the people who are convinced that hydroxychloroquine is an amazing cure that Big Pharma is suppressing will not be swayed by anything short of something that tells them exactly that.”

©2020 Bloomberg L.P.