Australasian Science: Australia's authority on science since 1938

The Tamiflu saga shows why all research data should be public

By Chris Del Mar

Attempts to evaluate whether the antiviral drug Tamiflu is effective have been stymied by lack of access to the data from clinical trials.

There’s a dispute going on at the moment, a war of words with lots of public relations manoeuvring. It’s all about the disclosure of trials undertaken in the past. And it is between Cochrane reviewers (of which I am one) and Big Pharma (in this case, Hoffman-La-Roche of Switzerland).

To understand the dispute, it’s necessary to go back a bit in history.

Tamiflu (oseltamivir) is a drug, cleverly-designed in 2000 to attack a molecule in the influenza virus. It prevents the influenza virus bursting infected human cells to spread more virus in the body.

It works rather like antibiotics on bacteria, except that it’s very specific to viruses or, rather, one virus exclusively – the one that causes influenza. It looked as if this class of drug was going to revolutionise treating this virus' infections, in much the same way that antibiotics have made treatment of once-dreaded bacterial infections routine.

Roche performed trials and the results were used to get approval from different regulatory authorities for it to be used routinely. The agencies include the US FDA (Federal Drug Administration), Europe’s EMA (European Medicines Authority), and Australia’s TGA (Therapeutic Drug Administration).

The trial reports supplied to regulatory authorities are called clinical study reports, and they are massive documents with added volumes of appendices, which set out every important detail of the trial.

The FDA approved Tamiflu for shortening the illness, but not for preventing complications. “Complications” means secondary (bacterial) infections, such as pneumonia. The EMA and TGA approved it for both indications.

This difference in approval is important.

The approval process is secret, ostensibly for “commercial interests”, although what these are at this stage of the drug’s development (testing how effective the drug is) remains unclear. Still, we often don’t know outside the closed doors what issues are raised in the clinical study reports.

Some of Roche’s trials were subsequently published in journals, and the drug was marketed reasonably successfully. It was extremely successful in Japan, where half of the world’s production was sold.

Enter the Cochrane Collaboration. This is a non-profit organisation that sets out to help clinicians and their patients decide which treatments are effective, and how effective.

It does this by systematically collecting all the trials done that answer a clinical question (such as “does Tamiflu reduce the symptoms of influenza?”), and then seeing if the results can be combined statistically to provide an overall answer.

Cochrane reviews are helpful to clinicians, especially when different trials on the same topic provide different answers. They help clinicians decide whether a treatment is likely to be effective for a particular patient. They also help health administrators decide whether to promote, or even subsidise that treatment.

This is called a “systematic review” and Cochrane reviews have the most rigorous process. It includes feedback, enabling anyone to critically comment on the review and how it was conducted.

A Cochrane review of neuraminidase inhibitors (the class of drugs that Tamiflu sits in) was conducted and published in 2006. It concluded that Tamiflu was effective at shortening influenza symptoms, and reducing complications.

But feedback from a Japanese paediatrician in 2009 highlighted the review’s uncritical acceptance of some published data incorporated in the process.

This led to a re-evaluation of the recommendations and the conclusion of the review. It now said that whether Tamiflu reduced or prevented complications of influenza was “uncertain”.

This was concordant with the FDA’s refusal to approve the drug for reducing complications.

We (the updated international team of Cochrane reviewers) wrote to the manufacturer to obtain the necessary data to resolve the issue. Long and complex interchanges that have now been made public took place.

We still haven’t received all the data we need, and in our next update, have resorted to information from the FDA and EMA obtained in a variety of ways. But inconsistencies in what was published and what we have, suggest there are still more data needed for answers to these questions.

The whole rigmarole has two implications.

First, we still remain in some doubt about the effectiveness of Tamiflu for preventing influenza complications. This matters because it was the prime reason the world armed itself with warehouses of Tamiflu under the threat of pandemic influenza in 2009.

Roche sold literally billions of dollars worth of Tamiflu for this singular reason. But has the world been sold a pup? Until the trials are openly released for scrutiny, it’s impossible for us, the world, the purchasers of this drug, to know.

Second, the implications are far beyond just Tamiflu. Until recently, the world had adopted the randomised controlled trial as the benchmark test for treatments. But there are clearly some problems with this.

One of these problems is the partial release of trial data (effected by selective publishing) that will help the sales of a commercial drug or device.

We’re concerned that this might have happened in this instance, even though the drug company in question has declared (several times, including originally in 2009) that it will provide all the data we need.

This kind of thing has happened before: drugs have had dangerous side effects suppressed by the manufacturers resulting in hundreds of patient deaths.

There are two ways this can be fixed.

Registering clinical trials needs to be made mandatory so we know they exist. This has progressed and there are many registers available for registering trials that can be searched.

And that all trial data is published needs to be made mandatory. That trials can be conducted on patients as guinea pigs – who sign up to them in the belief that this will somehow advance medical science and their fellow humans – but never see the light of day, seems to verge on the unethical. (Most patient trial data on Tamiflu have never been published).

There’s a campaign for this called Alltrials. Sign up as an individual. Get your institution to sign as well. This is important.

In the meantime, we’re still working on unravelling the Tamiflu mystery using sub-optimal information… Watch this space.

Professor Chris Del Mar is professor of public health at Bond University. This article was originally published at The Conversation.