Tug of war for antiviral drugs data.

Tom Jefferson can recite the details of every published study on the neuraminidase inhibitors oseltamivir (Tamiflu) and zanamivir (Relenza): date of publication, authors’ names, journal title, research question, methods, funder, and findings. But what has occupied the mind of the physician-academic from Cochrane’s Acute Respiratory Infections Group for the better part of five years is another detail about the science behind the influenza antivirals: that much of it has been missing or unpublished. Before he knew better, in 2006, Jefferson waded into the evidence and released a Cochrane systematic review about the neuraminidase inhibitors based on published research findings. He and his coauthors found that the drugs decreased the risk of influenza complications and hospital admissions in adults. By 2009, however, Jefferson’s opinion had started to shift. Europe was in the clutches of the influenza A/H1N1 panic, and he and his peers were asked to update their analysis of the flu drugs. Just as he was stepping into the research again, a letter arrived from a Japanese paediatrician, Keiji Hayashi. He was concerned that he could not verify the data that supported some of Jefferson’s conclusions about oseltamivir. In particular, he pointed out that the claims that the drug reduced secondary complications and hospital admissions were based on a 2003 analysis coauthored by Roche employees. Hayashi also noted that eight out of the 10 trials included in that analysis were never published in peer reviewed journals. In other words, Cochrane’s previous studies on the influenza treatments were based on an incomplete and possibly biased picture of the evidence. The Cochrane team didn’t have an answer for Hayashi. They didn’t even know about the existence of the unpublished research. “This is when I started thinking there is something wrong,” Jefferson now recalls. He felt duped and worried that his previous work was misleading physicians, patients, and public health authorities. “I don’t like being made a fool,” the former British military physician, told the BMJ. “I trusted literature. I trusted people who were doctors and researchers. I trusted the archive. I trusted Roche.” That trust evaporated. In December 2009, Jefferson and the Cochrane group published an updated review on neuraminidase inhibitors in the BMJ. This time, the researchers asked Roche for raw data from the unpublished trials. Roche, in exchange, asked them to sign a confidentiality agreement with a secrecy clause. The academics refused andwent ahead with their review, noting the inconsistencies in the evidence and omitting the studies based on unpublished trials. Without the raw data about oseltamivir, they wrote that the drug may work no better than aspirin. They also could not say whether it truly reduced the risk of serious complications and hospital admissions. Data promises