By Peter Doshi and Tom Jefferson

In the fall of 2009, at the height of fears over swine flu, our research group discovered that a majority of clinical trial data for the anti-influenza drug Tamiflu ― data that proved, according to its manufacturer, that the drug reduced the risk of hospitalization, serious complications and transmission ― were missing, unpublished and inaccessible to the research community. From what we could tell from the limited clinical data that had been published in medical journals, the country’s most widely used and heavily stockpiled influenza drug appeared no more effective than aspirin.

After we published this finding in the British Medical Journal at the end of that year, Tamiflu’s manufacturer, Roche, announced that it would release internal reports to back up its claims that the drug was effective in reducing the complications of influenza. Roche promised access to data from 10 clinical trials, 8 of which had not been published a decade after completion, representing more than 4,000 patients from every continent except Antarctica. Independent verification of the data seemed imminent. But more than two years later, and despite repeated requests, we have yet to receive even a single full trial report. Instead, the manufacturer released portions of the reports, most likely a very small percentage of the total pages. (One of us, Tom Jefferson, has been retained as an expert witness in a lawsuit relating to some of these issues.)

This is entirely within Roche’s rights. After all, regulators have never required drug or medical device manufacturers to share their data with independent researchers or academics. They are required to show the information only to the regulators themselves, who treat the data as secret.

Some may argue that, because the Food and Drug Administration approves drugs for the United States market based on these data, this is not a major cause for concern. But the actual use of drugs is often driven by assumptions about drug safety and effectiveness put forth by articles in peer-reviewed journals (sometimes written by doctors affiliated with the drug manufacturers) and clinical practice guidelines that can be entirely inconsistent with the F.D.A.’s assessments.

In the case of Tamiflu, some of these assumed properties led to stockpiling at great taxpayer expense ― more than $1.5 billion. The F.D.A. approved Tamiflu for the treatment of influenza (on the basis that it could reduce the duration of flu symptoms by about a day); not for the prevention of transmission. But other agencies are far more enthusiastic about Tamiflu’s benefits. The Centers for Disease Control and Prevention has argued that it reduces the risk of hospitalizations and serious complications like pneumonia, citing Roche-authored papers. The Department of Health and Human Services, also citing Roche, assumed in its national influenza pandemic plan that Tamiflu would cut complications. And the World Health Organization’s pandemic planning assumed that the drug would cut transmission of the virus. But here’s the rub: none of these organizations have vetted the original trial data.

The only agency in the United States that seems to have independently reviewed the original trial data never made these claims. The F.D.A.’s conclusion ― which it required Roche to print on Tamiflu’s product labeling ― is that “Tamiflu has not been shown to prevent” complications like serious bacterial infections (for instance, pneumonia). It seems that federal agencies like the C.D.C. and H.H.S., instead of conducting an independent evaluation of Tamiflu, advocated stockpiling by referencing claims in journal publications written by the drug’s manufacturer, ignoring the F.D.A.’s assessment that those very claims were unproven.

Why would they do this? Unwarranted trust in the peer-review process of medical journals probably has something to do with it. So, too, does wishful thinking; lacking good alternatives, it’s tempting to hope that the drug we have works wonders―not to mention the pervasive (but mistaken) belief that something is always better than nothing. And it’s important to remember that correcting the statements of medical journals or public health agencies falls outside the F.D.A.’s jurisdiction ― when it comes to drugs, the F.D.A. is responsible for regulating industry, not other government agencies.

But this is no way for supposedly evidence-based decision making to work, and the F.D.A. could do much more. As a result of new freedom of information policies in Europe, the Continent’s version of the F.D.A., the European Medicines Agency, has released 22,000 more pages of Roche’s Tamiflu trial reports. But even this represents an incomplete picture, as the most detailed portions of the reports are not in the European drug regulator’s files.

Nevertheless, the data point to a drug of minimal benefit. In accordance with the F.D.A.’s findings, it appears to shave a day off the duration of influenza symptoms, but we found no decrease in risk of hospitalization and no evidence that it could stop the spread of the virus. More worrisome, we found suggestive evidence that Tamiflu interfered with the body’s ability to produce antibodies against influenza ― which could affect the body’s response to influenza vaccine and its ability to fight off future influenza infections. But to do a complete analysis, including evaluating Tamiflu’s potential harms, we need the remainder of the data ― the full “clinical study report” ― promised by Roche, but never delivered for a variety of shifting reasons that we have recently detailed in an article in PLoS Medicine.

In response to our conclusions, which we published in January, the C.D.C. defended its stance by once again pointing to Roche’s analyses. This is not the way medical science should progress. Data secrecy is a disservice to those who volunteer their bodies for clinical trials, and is dangerous to those being asked to swallow approved medicines. Governments need to become better stewards of the scientific process. The European regulator’s announced intention to release clinical study reports after it finishes reviewing a manufacturer’s application is an important precedent. Their position―that “clinical trial data should not be considered commercial confidential information”―is a bold step in the right direction.  But the F.D.A. ― guardian of arguably more trial data than any other entity in the world ― appears stuck in the era of data secrecy.

We should not have to wait for patients to be hurt by the medications they take, as recently happened with the diabetes drug Avandia, before reviewing this wealth of data.

Peter Doshi is a postdoctoral fellow in comparative effectiveness research at Johns Hopkins University School of Medicine. Tom Jefferson is a Rome-based independent epidemiologist with the Cochrane Collaboration, an international nonprofit research organization.

A version of this op-ed originally appeared in the New York Times on April 10, 2012.