By Tom Chivers Science Last updated: April 11th, 2014

Flu is actually a pretty nasty disease. People say they’ve got “the flu” when they’ve really only got a cold, but a genuine bout of influenza knocks you off your feet for several days, and is a major killer in the elderly; an outbreak in 2011 killed around 600 people. But don’t worry! Since 2006, during the avian flu scare, the Government has spent £424 million stockpiling a drug called Tamiflu.

Unfortunately for the Government, and for elderly people, and for people who don’t want to spend a week in bed aching and sweating and vomiting, it seems that Tamiflu doesn’t work very well.

The Cochrane Library, the great centre for epidemiology and public health data, has carried out a huge meta-analysis, a study combining the data from dozens of smaller trials, into Tamiflu and another “neuraminidase inhibitor”, Relenza. Cochrane Reviews are hugely authoritative works, a sort of gold standard if you’ll forgive the cliché. And, five years after the UK and US began spending billions to buy millions upon millions of doses, the Cochrane Library has reported.

On average, flu symptoms last for seven days. With treatment with a neuraminidase inhibitor, they last, on average, for 6.3 days. Consider the front page held.

It’s worse than that for Tamiflu and Relenza. Even that limited improvement was unclear in children, and “there was no evidence of a reduction in hospitalisations or serious influenza complications (confirmed pneumonia, bronchitis, sinusitis or ear infection) in either adults or children,” says the Cochrane Review. And the study found an increase in cases of vomiting and nausea and an increase in risk of psychiatric events. Roughly speaking, for every 100 people treated with the drug, three or four people would suffer nausea or vomiting who wouldn’t have otherwise, and one extra person can be expected to suffer an adverse psychiatric reaction. (This is known as the “number needed to harm“.)

And yet this drug has not only been on the market for years, it’s been actively stockpiled by the NHS as a frontline defence against flu pandemics. How on earth, you might reasonably ask, has a drug of such startlingly limited use managed to get to this position?

The answer is depressingly simple: Roche and GlaxoSmithKline didn’t release their data. The Cochrane Library has, after years of trying, got hold of 20 Roche studies into Tamiflu and 24 GSK studies into Relenza. Fiona Godlee, the BMJ’s editor, says that “This review is the result of many years of struggles to access and use trial data, which was previously unpublished and even hidden from view.” But the data had already existed, and was simply sitting in Roche and GSK’s vaults. None of the information used to create this review is, in itself, new. It’s all been around for years. The only thing that’s new is that people outside the companies that make the drugs have access to it. And it’s taken the Cochrane Library years of arguing with Roche to get it; the company, at first, insisted on confidentiality agreements, and then only gave snippets of data. Credit to them, I suppose, for releasing it at last.

Of course it’s appalling that so much money has been wasted on a drug that barely works. But the real scandal, as Ben Goldacre, a doctor and journalist behind the AllTrials campaign for transparency in trial data, says, is that “Roche broke no law by withholding vital information on how well its drug works“. In fact, as he points out, it’s actually just Roche’s bad luck that they’re the ones caught out by this; “the methods and results of clinical trials on the drugs we use today are still routinely and legally being withheld from doctors, researchers and patients”, and it could have been any one of dozens of other treatments into dozens of other diseases which happened to be the one which was caught out.

I spoke briefly to one of the Cochrane authors, and he mentioned two other drugs that made it to patients before dangerous side effects were discovered. In one case, an antidepressant called paroxetine (marketed as Seroxat), caused an increase in suicides among adolescents, a fact that was only discovered when GSK was forced to release its data in a lawsuit in 2007. In another case, rofecoxib (aka Vioxx), an anti-inflammatory used to treat rheumatoid arthritis, was pulled from the shelves in 2004, after its manufacturer Merck was shown to have misrepresented trial data showing that the drug caused heart attacks. A Lancet study found that between 90,000 and 140,000 people died unnecessarily after taking it.

Transparency is important. The Cochrane Library is brilliant not only because its reviews are comprehensive, but because they are made available, in full, for free, online – I found the full Tamiflu paper in a couple of minutes – so that anyone, any company or academic or regulatory body, can check their working. Medical science – all science – relies on results being replicated; if I say that I can create cold fusion in my garage but won’t show my working, you will be unlikely to believe me. More relevantly, if I carry out 10 studies into my new wonder drug, and seven of them come back saying “no result” and three of them come back saying “some result”, and I only publish the data from the three that are positive, I immediately skew the available information in the public sphere, as clearly as if I’d rolled 10 dice and only released the times I rolled a six.

But in pharmaceutical research, there is – as yet – no requirement to release the data about the times you rolled a one. The AllTrials campaign wants every single medical trial to be registered and reported, so that the dice are no longer loaded. Otherwise, scandals like the Tamiflu waste of money will be repeated – and worse, patients will die unnecessarily – and the companies behind them will have broken no law.

More on Tamiflu

• When drug companies don’t publish results, people die
• Swine  flu and Big Pharma: why is there no scandal?
• The cost of the swine flu scare