Drug companies' trials under scrutiny


A review of studies into the effectiveness of anti-flu drugs Tamiflu and Relenza raises some awkward questions about pharmaceutical companies' methods. By Norman Swan.  


Istock & Getty Images

Can we fully trust the information that drug companies give us? There’s long been a suspicion that they gild the lily; that the claims made for the effectiveness of drugs in prestigious medical journals are not backed up by the raw data.

This question was brought to a head by a recent bout of flu – H1N1 to be precise. Back in 2009 authorities in many countries feared it would cause a disastrous pandemic. Nations that could afford to stockpiled expensive anti-influenza medications called neuraminidase inhibitors in the belief that the drugs reduced the incidence of life-threatening pneumonia, shortened the illness, and slowed the spread of the infection. The most famous of these inhibitors is oseltamivir (which goes under the brand name Tamiflu), followed by Relenza (which was developed in Australia).

The Australian and British governments, despite having bought their stockpiles, wanted to be sure about the drugs’ effectiveness, so they commissioned a systematic review of the available studies from a group in the Cochrane Collaboration – a global network of clinicians, statisticians, epidemiologists and consumers who use sophisticated analytic techniques to assess the quality of clinical trials, combine the best data, and sum up whether interventions work.

Four years ago they did a first pass and were shocked by the poor quality and questionable independence of the drug companies’ trials.

Their suspicions were aroused because the US Food and Drug Administration (FDA) had restricted what the manufacturers were allowed to claim, presumably because the FDA had been privy to the raw patient data that sat in the company vaults and might not have seen the light of day in published papers.

For all of the Tamiflu trials and several of the Relenza studies, the data on the final outcomes of the flu treatments were incomplete.

The Cochrane researchers requested the same data from Roche, the owner of Tamiflu, and so began a four-year row involving an international shaming campaign to get the data. Once the reviewers did, they performed a second analysis, which was published this April.

What emerged was troubling to say the least.

The company drug trials are randomised, double blind, placebo-controlled trials. If done correctly they are the best method we have of eliminating bias. Each word points to a crucial component. Randomisation means the people in the active group should be exactly the same as the people in the placebo group for things like age, gender, education, illness, lifestyle etc. Double blind means that neither the patient nor the treating clinician knows which group the patient is in, and crucial to that is the placebo medication, which must be identical to the drug in every respect except that it doesn’t contain the active ingredient.

In at least 11 of the 20 trials of Tamiflu, the placebo appeared different from the drug, which could have compromised the “blinding”, and there were doubts about some of the Relenza placebos too. People receiving antibiotics, a factor that could influence the course of the illness, weren’t excluded and the numbers of participants were sometimes too small to be able to draw conclusions. For all of the Tamiflu trials and several of the Relenza studies, the data on the final outcomes of the flu treatments, including important things like hospitalisations, were incomplete. It was hard to match the stories of individual patients to what had been published and side effects were poorly defined and recorded. There was selective reporting. The pharmaceutical companies chose what data to release and the objectives of the trials shifted frequently after the trials started. These are big no-nos, as if we hadn’t enough no-nos already.

The net finding? Neither of the medications reduced hospitalisations, the spread of infection or the number of cases of proven pneumonia.

And to top that, Tamiflu in particular had under-reported harms such as psychosis, depression, anxiety, kidney problems, nausea, vomiting, and high blood sugar levels. According to the researchers, if you gave Tamiflu to a million healthy people for influenza prevention, 200,000 would experience nausea and vomiting, 150,000 would report headaches, 10,000 would have psychiatric events, 6,000 kidney problems, and 4,000 high blood sugar. That’s a lot of people suffering for little or no benefit. Relenza had fewer adverse events, probably because it’s inhaled and less enters the body. For the sick and elderly, the reviewers reckoned the rate of side effects would be higher.

The huge and scary question is how many medications on the market today would survive a similar analysis? How much data have been withheld (which by the way is not illegal in most jurisdictions), wasting money and putting patients at risk? The task of finding out would be huge. The Tamiflu exercise, for example, interrogated 150,000 pages of data.

Many critics of the industry are now calling for a systematic response on the part of authorities and compulsory availability of patient data for bona fide researchers, not to mention standardisation of what is given to regulatory bodies around the world. The irony was that the FDA was aware the data on Tamiflu were thin but that didn’t stop another part of the US Government buying massive supplies, turning the antiviral into a blockbuster.

Most of us expect that regulatory authorities look at all the data when making a decision. They don’t because there isn’t the staff, the funding or the government support to insist on it. One solution would be international cooperation, sharing the work, so that new drugs receive the attention they deserve. But that would mean regulatory agencies accepting the findings of others.

Can egos and empires and vested interests stand back to protect public safety and the public purse?

Latest Stories
MoreMore Articles