Forget everything you've heard about the benefits of Tamiflu, the Roche corporation's blockbuster influenza drug. A new analysis that includes previously-unpublished trials offers a fresh, and very unflattering, perspective.
Selective publication, in which some trials of a drug are released publicly and others are not, is widely considered unethical. In addition to producing a keyhole view of the data, non-publication misleads study subjects, who believe that their participation is for the broad advancement of medical knowledge. In the mid-1990s, however, Roche commissioned and funded eight Tamiflu trials (that we know of) and then stored them in a proprietary database. Other than a 2003 paper that offered a limited, and highly favorable, view of the results (it was authored by company employees), the scientific community has had little opportunity to examine the trials.
Last year, still under pressure from journal editors and independent researchers, Roche granted access to a group from the University of Georgia led by Dr. Mark Ebell, a respected health evidence expert. Published in the journal Family Practice in September 2012, the team's analysis combines data from the eight unpublished trials with data from three published ones and gives a far less obstructed view of Tamiflu's effects.
Two findings are notable. The drug, it turns out, does not reduce the chance of serious illnesses like pneumonia, despite earlier claims by Roche. This was the question independent researchers were seeking to answer in 2009 when they were denied access to the Roche database for the first time.
The finding is, in a sense, unsurprising. If a car company refused access to its crash tests results, what might you think they showed? Still, it is no small matter. When Tamiflu was approved in 1999, I was a resident in training. The pricey drug cut flu symptoms like fever and body aches by less than a day when compared to a placebo. Worse yet, it caused nausea and vomiting, and only worked if it was given in the first two days of the illness. Few if any of my peers prescribed or recommended Tamiflu.
Things changed, however, after 9/11 and the anthrax attack. A new narrative emerged for Tamiflu: In the event of bioterrorism the drug could reduce serious illness from influenza infections. This narrative relied on the 2003 study. In the years that followed, with headliners like SARS, Avian flu, and swine flu, the narrative took. Officials at the Department of Defense spent billions to stockpile Tamiflu, and the Centers for Disease Control and Prevention, along with state health departments, began recommending the drug. This was in spite of a statement from the Food and Drug Administration, the only group other than Roche to have seen the complete data, that Tamiflu had no effect on serious illness.
The question now is whether the narrative has left the station. Last week, for instance, the journal Pediatrics published a methodologically anemic study claiming that children in California who died from influenza might have survived had they been given the drug. The difference between groups (2 percent) was fatuous, and the California study is trumped by the new review of trials that finds no impact on the development of serious illness. The compelling narrative of a life saving treatment for influenza could take years to puncture and deflate.
Interestingly, there is another finding from the new paper that may be even more powerful: Two of the unpublished studies enrolled the elderly and chronically ill, those with the highest risk for severe infections and complications from influenza. Despite including over a thousand such patients, the trials not only failed to show any decrease in serious illness, in this group Tamiflu also had no impact on symptoms.
There will be more to come. Earlier this year the British Medical Journal and two tenacious researchers pressured the company into releasing its original trial documents, an even deeper look. But while awaiting further enlightenment, we appear to have uncovered two critical facts. First, Tamiflu does not prevent serious illness and should not be recommended for this use. Second, the elderly and chronically ill seem to gain nothing from the drug, other than side effects. The CDC and state health departments should amend their recommendations to reflect this new scientific knowledge.