It mainly found the medicines had few if any beneficial effects, but did have previously dismissed or overlooked adverse side effects.

"Remember, the idea of a drug is that the benefits should exceed the harms," said Carl Heneghan, a lead investigator with the Cochrane review and a professor of evidence-based medicine at Oxford University. "So if you can't find any benefits, that accentuates the harms."

But Roche, which has been under fire for several years over its refusal to allow the Cochrane team unrestricted access to Tamiflu data, rejected the findings, saying it "fundamentally disagrees with the overall conclusions" of their study.

"We firmly stand by the quality and integrity of our data, reflected in decisions reached by 100 regulators across the world and subsequent real-world evidence demonstrating that Tamiflu is an effective medicine in the treatment and prevention of influenza," it said in a statement.

Tamiflu sales hit almost US$3 billion in 2009, mostly due to its use in the H1N1 flu pandemic, but they have since declined.

The drug is approved by regulators worldwide and is stockpiled in preparation for a potential global flu outbreak. It is also on the World Health Organisation's "essential medicines" list.

The United States has spent more than US$1.3 billion buying a strategic reserve of antivirals including Tamiflu, while the British government has spent almost £424 million (HK$5.5 billion) on a stockpile of some 40 million Tamiflu doses. In 2005, Hong Kong announced it would spend HK$254 million on a stockpile of 20 million Relenza and Tamiflu tablets, equivalent to two million doses - enough to cover about 30 per cent of Hong Kong's population of about seven million.

Heneghan's team said its analysis was the first based on full data: from 20 trials of Tamiflu, known generically as oseltamivir, and 26 trials of Relenza, also known as zanamivir.

At a presentation in London, Heneghan said the money spent on stockpiles had been "thrown down the drain" because until now, the full data had not been seen by regulators, governments, doctors or patients.

"The original evidence presented to government agencies around the world was incomplete," said Fiona Godlee, editor of the , which has spearheaded a four-year campaign to force Roche to reveal all its Tamiflu data. "And when they (the Cochrane review team) eventually received the full information on these drugs, the complete evidence gives a very much less positive picture."

Adding more heat to the row, the European Medicines Agency (EMA) - which approved the drug for sale in Europe - disputed the claim they had not seen all the Tamiflu data.

Enrica Alteri, head of medicines evaluation, said the EMA had seen and reviewed all 20 studies referred to in the review. She said the latest analysis did not raise any new concerns or alter the agency's assessment that the benefits of Tamiflu outweighed its risks.

The Cochrane review found that compared with a placebo, Tamiflu led to a quicker alleviation of flu-like symptoms by a factor of around half a day (down from seven days to 6.3 days) in adults, but the effect in children was more uncertain.

There was no evidence of a reduction in hospital admissions or in serious flu complications such as confirmed pneumonia, bronchitis, sinusitis or ear infections in either adults or children, Heneghan's team said. Tamiflu also increased the risk of nausea and vomiting, and mood changes when used as a preventative drug in uninfected people.

Godlee described the battle with Roche as a "really lengthy cat-and-mouse, , bizarre experience of trying to get data on a drug which governments around the world were busy buying, stockpiling and spending billions of dollars on."