“Tamiflu maker accused of secrecy over trial data,” reported The Independent today, while The Daily Telegraph said that scientists have challenged Roche, the manufacturer of the antiviral, to “prove Tamiflu works”. While these are disturbing headlines, they do not necessarily mean that the drug is ineffective or even harmful, as some newspapers seem to imply.
These headlines have come about following the publication of a systematic review into the effectiveness of Tamiflu (oseltamivir) and Relenza (zanamivir) in preventing and treating flu in adults and children. These drugs are from a group called neuraminidase inhibitors (NIs), which are thought to help reduce symptoms of flu. However, the evidence behind their effectiveness is a source of continued debate.
This story is particularly important because the UK government has spent millions of pounds stockpiling Tamiflu to protect against the threat of a potential flu pandemic, as have many other countries.
The review of Relenza was postponed due to new information about how the drug affected individual patients being made available by the manufacturer (GlaxoSmithKline). The results of this are eagerly awaited. The review on Tamiflu was incomplete because of difficulties obtaining sufficiently detailed information from the manufacturer (Roche).
Overall, the review included 25 studies, but had to exclude 42 relevant studies due to the lack of patient information or unresolved problems in the data. Not all of these related to Tamiflu and its manufacturer, Roche. Excluding this amount of information significantly affected the results of the review and prevented any realistic conclusions from being drawn. Patients volunteering for these trials may be disappointed to hear that so far their data has not been made available for open analysis.
The debate on whether Tamiflu is effective continues. The simplest way to resolve it would be to allow independent reviewers to access the full existing results of studies into it.
Where did the story come from?
The study was carried out by researchers from The Cochrane Collaboration and was funded by a grant from the National Institute for Health Research UK. It was published in the peer-reviewed medical journal The Cochrane Library.
The majority of media reporting was balanced. However, the media mainly focused on the fact that Roche, the manufacturer of Tamiflu, did not provide sufficient data for the review to be effective, which hampered the researchers' ability to make worthwhile conclusions.
What kind of research was this?
This study was a systematic review aiming to identify all high-quality clinical research studies that looked at how effective NIs were at preventing and treating flu in healthy adults and children. Two NIs were assessed: oseltamivir (brand name Tamiflu, manufactured by Roche) and zanamivir (brand name Relenza, manufactured by GlaxoSmithKline).
Preparing for potential pandemic flu outbreaks are a high public health priority for national governments. NIs are a class of drug thought to help prevent and reduce symptoms of flu but the evidence behind their effectiveness is a source of continued debate. Despite this, many governments including the UK have stockpiled drugs of this kind. The Department of Health has purchased large quantities of Tamiflu to prepare for a potential flu pandemic.
In a previous review, The Cochrane Collaboration reported substantial problems in assessing the effectiveness of NIs. They said that the drug manufacturers had never published large amounts of the results of trials of the drugs. The current review was done because the authors became aware of a large number of unpublished results of NIs and they wanted to update previous reviews with new results as well as previously unpublished results.
Systematic reviews are the best way to assess the effectiveness of NIs in combating flu because they summarise information from all available research (ideally large clinical trials). However, the quality of this type of review relies heavily on identifying and including all relevant information on the topic. If some studies or results are not included, this can bias the results of the review.
What did the research involve?
The researchers comprehensively searched the scientific literature for published and unpublished sources of information on the effectiveness of the two NIs up to 12 April 2011. They selected information from placebo-controlled randomised trials conducted in people of any age who had:
- suspected flu
- been exposed to flu in their homes or local community
They also included information and reviews from manufacturers and drug regulators.
When discrepancies in data were found between different sources, the review authors contacted the relevant manufacturers for explanations and further information. Where possible, the authors combined the results of the individual studies identified to produce a single combined estimate of the effect of the drugs.
What were the basic results?
The main results were as follows:
- Twenty-five studies were included in the final review including 15 on Tamiflu and 10 on Relenza. All studies identified were funded by the drug manufacturers.
- The information from a further 42 studies could not be used due to insufficient information or unresolved discrepancies in their data.
- An evidence review on Relenza was postponed after the manufacturer offered information about its effects on individual patients to aid the review.
- Tamiflu reduced the time to first alleviation of symptoms in people with flu by around 21 hours (range: 29.5 to 12.9 hours) compared to those given a placebo. On average, the placebo groups’ symptoms started to improve at around 160 hours (just under seven days) whereas those receiving Tamiflu improved an average of 21 hours (just less than one day) earlier. This was based on five studies.
- There was no evidence that Tamiflu reduced the chance of a person with flu needing to be hospitalised compared with placebo. This was based on the combined results of seven studies. Hospitalisation due to flu-like symptoms was rare and occurred in 0.84% of the placebo group.
- People randomised to receive Tamiflu were 17% (95% confidence interval 0.73 to 0.94) less likely to be diagnosed with flu compared to a placebo group. This was based on combined results of eight studies. Trials assessing Relenza showed no evidence of this.
- Limitations in the design, conduct and reporting of trials of Tamiflu meant the reviewers were unable to assess the effect of Tamiflu on complications of flu or flu transmission.
How did the researchers interpret the results?
The researchers concluded there was a high risk of publication bias and reporting bias in trials assessing Tamiflu. Publication and reporting bias can occur when not all the results from available trials are included in a systematic review or when not all trials on the topic are published. This means that some results are excluded from the review which, had they been included, could have influenced the overall conclusions.
The reviewers stated that they were “unable to obtain the full set of clinical study reports or obtain veriﬁcation of data from the manufacturer of oseltamivir [Tamiflu] (Roche) despite ﬁve requests between June 2010 and February 2011 [eight months].”
They also stated that they “expect full clinical study reports containing study protocol, reporting analysis plan, statistical analysis plan and individual patient data [from Roche] to clarify outstanding issues. These full clinical study reports are at present unavailable to us.”
The researchers concluded that the limited evidence available to them suggests Tamiflu can reduce symptoms of flu but they were unable to assess its effects on complications or transmission.
This systematic review aimed to assess comprehensively the effect of NIs including Tamiflu and Relenza on the prevention and treatment of flu in healthy adults and children by including results known to have been missed in previous reviews. In this sense, the review failed to meet its aims but for two distinct reasons.
The evidence review of Relenza was suspended because new information about its effects on individual patients became available. The results of this are eagerly awaited.
The review of Tamiflu was incomplete because of difficulties in obtaining sufficiently detailed information from the manufacturer.
The systematic review included 25 studies in its final analysis, but had to exclude 42 relevant studies. By excluding these relevant studies, important information that could influence the conclusions may have been missed. This led the authors to conclude that their results had high levels of publication and reporting bias. Given the high-profile nature of Tamiflu, the conclusions of the review should be treated with great caution. It may be prudent to postpone the judgement on whether Tamiflu is effective until after the outstanding trial information has been obtained and incorporated into the review.
The Daily Telegraph and other news sources have quoted a Roche spokesman as having said that Roche made full clinical study data [on Tamiflu] available to health authorities around the world as part of the licensing process.
The debate on whether Tamiflu is effective continues. The simplest way to resolve it would be to allow independent reviewers to access the full existing results of studies into it. These are currently unavailable to them, but the reasons for this are unclear.