This past weekend, Donald Trump used his daily White House coronavirus briefings to again urge Americans to take hydroxychloroquine, an anti-malaria drug that has not been shown to be safe or effective against Covid-19.
“What do you have to lose? Take it,” the president said on Saturday as he boasted that the US had amassed 29m doses of the drug. On Sunday, facing questions from the press about his aggressive promotion of an unproven treatment, he argued against waiting for the completion of clinical trials. “In France, they had a very good test,” he said. “But we don’t have time to go and say, ‘Gee, let’s take a couple of years and test it out, and let’s go and test with the test tubes and the laboratories.’”
Meanwhile, Dr Anthony Fauci, the country’s top infectious disease doctor, has repeatedly warned that there is no conclusive evidence to support using the drug. Asked whether it should be considered a treatment for Covid-19, he said on 24 March: “The answer is no.”
The story of how hydroxychloroquine was anointed the Trump administration’s miracle drug for the coronavirus pandemic is a distinctly modern tale of misinformation within a global information ecosystem beset by widespread uncertainty, fear, media fragmentation and hyper-partisanship. Belief in the drug’s potential to cure patients infected with the virus followed an extraordinary trajectory from a small study conducted in France (Trump’s “very good test”) to Silicon Valley social media influencers, Fox News and the largest bully pulpit: the White House.
But it’s also a story as old as medicine itself. When an epidemic killed thousands in ancient Rome, said Aaron Shakow, a research associate at Harvard Medical School and historian of medicine, the chief physician of the emperor Nero circulated a recipe for an old miracle cure.
“It was an attempt by Nero to sustain his legitimacy in the midst of this catastrophic event,” Shakow said. “Epidemics are dangerous to rulers.”
A deeply flawed study
In early March, as the coronavirus pandemic accelerated its spread around the globe, a group of scientists in Marseille, France, launched an experiment to see whether hydroxychloroquine, a well-known old malaria drug, could be what everyone was searching for: a cure.
Most small scientific studies live and die within the rarified domain of academic journals, but the French trial had a much more auspicious debut. Before the study was even published, in the International Journal of Antimicrobial Agents (IJAA), a lawyer falsely claiming an affiliation with Stanford University appeared on Fox News’s Tucker Carlson Tonight to declare the results: a “100% cure rate against coronavirus”. From Fox News, it was only a matter of time (hours, in fact) before the drug was being hailed as a “game changer” by the president of the United States.
Trump made his first endorsement of hydroxychloroquine on 19 March. Export controls, shortages, overdoses and scientific recriminations rapidly ensued, but the controversy could not extinguish the power of presidentially endorsed hope. Across the globe and throughout diverse communities on the internet, hydroxychloroquine had been anointed the miracle cure for Covid-19.
The only problem? The study that all this fervid hope is based on doesn’t show what its authors claim it does.
The gold standard for a clinical trial is a double-blinded, randomized controlled trial (RCT). What this means in plain English is that the study has been designed to reduce biases that would render its results meaningless. Neither the physician nor the patients knows whether they received the drug (“double-blinded”), a safeguard that reduces the possibility that the doctor will treat the two groups differently. The researchers also do not get to choose which patients go into which group (“randomized”) and the makeup of the two groups is roughly equivalent (“controlled”).
The French hydroxychloroquine study did not follow any of these rules.
The treatment group and the control group were drawn from separate populations: the treatment group were all patients at the institution where the researchers worked, the Méditerranée Infection University Hospital Institute in Marseille, while the control patients came from other hospitals in the south of France. The treatment group (mean age 51.2) was significantly older than the control group (mean age 37.3), introducing another variable that could undermine the meaning of the results. The study was “open label”, meaning the physicians and patients knew which treatment they were receiving. The French researchers also treated some but not all of the treatment group patients with azithromycin, a common antibiotic, another complicating factor that was not randomized.
But even more important than these shortcomings in the design of the study is how the researchers chose to measure and report their results. Forty-two patients were initially included in the study. Three were transferred to the intensive care unit; one died, one left the hospital, and one stopped taking the treatment due to nausea. The other 36 eventually recovered, and those who received the drug cleared the virus from the system faster than those who did not.
If you had only heard about this study from the Fox News assertion of a “100% cure rate”, you might assume that the four patients with poor clinical outcomes (the three ICU visits and one death) had been unlucky enough to be in the group that did not receive the “cure”.
And yet, those four patients, as well as the patient with nausea and the one who left the hospital early, were all part of the treatment group. They were excluded from the topline results of the study because of the way that the researchers chose to measure and report the results: strictly based on the measurable presence of viruses in nasal swabs taken each day of the study. Since the patients were in the ICU or dead, their samples could not be taken and they were left out of the final analysis. Based on the nasal swabs of just the 36 patients who completed the study, those who received the drug cleared the virus from their systems faster than those who did not.