Tamiflu is stockpiled globally as a defence against pandemic influenza … but does it work?

Billions of dollars have been spent, but the safety and effectiveness of Roche’s drug remain uncertain. Tony Hisgett

A Cochrane review published today has again raised questions about the efficacy of Tamiflu, the antiviral drug that has been stockpiled by countries across the world as the first line of defence against pandemic influenza. Tamiflu is on the World Health Organisation’s List of Essential Medicines but reviewers have found inconsistencies with published reports and possible under-reporting of side effects.

Professor Chris Del Mar, one of the review’s authors, spoke to us about the review’s findings and their implications.

On the face of it, this research is just an update of a Cochrane review about the effectiveness against influenza of a group of drugs called neuraminidase inhibitors. These drugs include the very famous oseltamivir, also known by its trade name of Tamiflu, and zanamivir, which is also know as Relenza.

These are the two most famous neuraminidase inhibitors but there are other drugs in this group as well. Both work on the same principle, but there are some important differences. The main one of these is that oseltamivir is taken orally or intravenously whereas zanamivir has to be breathed in, and that’s made it less attractive because people who are very sick with influenza often have trouble breathing so they might not be able to get the stuff down.

Neuraminidase inhibitors are antiviral drugs, specific for influenza and they work by inhibiting an enzyme that the virus has the host cell generate, which is important for the virus to escape from the cell it invades. The importance of research comes from the fact that the world has stockpiled billions of dollars worth of Tamiflu, made by the pharmaceutical and diagnostic manufacturer Roche.

The reason there’s such an interest in this drug is because it was stockpiled after claims by its manufacturer that it not only relieves symptoms but also reduces the complications of influenza. These complications include those that are the reasons people end up in hospital, such as developing secondary infections like pneumonia.

So stockpiling Tamiflu seemed to be an important public health intervention. It could be very important in case of a pandemic where large sections of the population gets infected with influenza. And that’s why Tamiflu got distributed and stored for such an event.

Then of course we had the avian flu scare and, following that, the swine flu scare. Neither of these were as severe as had been anticipated but they showed that there were good reasons to have such a drug around.

Cochrane reviews

Cochrane reviews work in the following way: world literature is reviewed and systematically analysed in a special statistical way to combine different studies and come up with a summary result.

The first review talked about how the complications from flu infection could be reduced by using neuraminidase inhibitors. We use an open process at Cochrane and someone wrote in and said that we didn’t do the first part of the analysis properly because we relied on secondary data, and we should get the primary data.

We took that on board and discovered the primary data weren’t available. We wrote to the author of the secondary review (Professor Kaiser Laurent from Geneva) that we took the data from – all quite proper – and asked, please can we have the primary data because we’ve had this criticism and we need to look at it and analyse it carefully.

He told us he didn’t have it, which was unexpected. It was extraordinary that he didn’t have the data, and he referred us to Roche.

Roche weren’t particularly helpful to us. That made us start wondering: where are these data? We realised that some of the data used for the assertion about oseltamivir reducing influenza’s complications belonged to a set of trials that had never been published. They were only published in this secondary form, from which we couldn’t look at the primary data.

So we asked Roche for the primary data and they said: “yes, we will give them to you,” but they’ve never given us the full data we would like.

From this point, we began to wonder whether the data that weren’t published would give as optimistic a result about the drug’s effect on complications as those that were published.

In our subsequent review, two years ago, we said we were uncertain about the effect. Since then, we’ve been able to get hold of some of the data Roche wouldn’t give us from regulatory authorities, specifically the FDA (USA) and EMA (Europe). So we’ve looked at some of the data Roche did give us and some they didn’t.

These authorities are the Federal Drug Administration (FDA) in the United States, and the European Medicines Agency (EMA), the Australian equivalent of whom is the Therapeutic Goods Administration (TGA). They had more information about the performance of the drug in trial evaluation because it’s required before they approve the drug. All drugs have to be approved for use as a drug for specific indications.

We went to the FDA and the EMA and they supplied us with some of the information, not all of what we wanted, perhaps because they themselves didn’t have it all. So we managed to get hold of some of the information that way.

And this review that we’ve published today is the result of the analysis from those data. It was a very difficult analysis because we didn’t have complete information and there was an awful lot of it. Even though it wasn’t complete, there were thousands and thousands of pages of reports, which we’ve been going through.

A clouded diagnosis?

We’re still not sure Tamiflu is effective for the complications of influenza. It looks to us as if it is reasonably effective at reducing symptoms – a little bit – but that in itself isn’t reason enough to have nations around the world stockpile billions of dollars worth of the drug as had happened.

In terms of its effectiveness in reducing symptoms, when you start thinking of a disease that gets better by itself, the difference between being less ill (“cured”) and feeling less ill is theoretical.

What’s more, there has been some criticism of this drug regarding its side effects. In Japan, there have been some reports of children, in particular, and adolescents who became psychiatrically very unwell after using it.

It’s very difficult to separate out an adverse effect like that as something that is the consequence of the drug, or something that this consequent on the illness that the drug is being used for. That’s always difficult to sort out but we’ve started to look at some of the data to try to work it out and we’ve got a new plan of analysis where we might be able to disentangle that.

At the moment there are certainly some indications that there are more adverse events being reported in the unpublished data while published summaries say they don’t exist. That is, some published reports declare no adverse effects while we’ve found some unpublished data reporting adverse effects, and some of them are attributed to oseltamivir itself.

Roche has funded nearly every single study on this drug; they are expensive studies to perform so they’re nearly always funded by the drug industry. Not only have they not been forthcoming with the data, they have never told us good reasons for not sharing it.

The company talks in vague terms about commercial-in-confidence reasons. That’s a very interesting argument to run because, although we think there are some aspects of the drug’s development that are commercial in confidence – particularly to do with its development in the earlier stages of a drug’s development when you could imagine competitors might get an advantage if they could get access to the information – the effectiveness and efficacy of the stuff, how well it performs in trials, is something that should be publicly available. This information should be in the public domain transparently because it’s to do with the public good.

In terms of pandemic preparation, having access to these data is about making sure that we’re not wasting our money on something. That if we’ve got a drug that’s not effective, that isn’t effective in reducing the spread of the disease, which is the most important thing in a pandemic, or in saving lives or saving people from having to go to hospital, then it’s absolutely vital that we have that information.

If it’s not effective on these important outcomes, it may be better for us to resort to something we do know works, like barrier methods – wearing masks, washing your hands a lot and quarantining people.