By-the-by I don’t really understand why the Guardian subs gave this piece, about how Tamiflu isn’t so great, a headline saying “the drugs do work”. I mean they kind of do work a bit, and we don’t know if they do in a pandemic since they’ve not been tested in those circumstances (which probably won’t come to pass) but we hope they will and so they’re recommended.

Ben Goldacre

The Guardian

Saturday 2 May 2009

Look I don’t want to freak you out, since Tamiflu is the one thing which everyone believes will save us from Parmageddon, but I’ve been reading through the published trial data on the drug, and I’m not sure it’s all that great.The Cochrane Library is one of the greatest inventions of modern humankind. It’s all very well to do a trial, or lots of little trials, but one trial, simply by chance, might give a false negative, incorrectly missing a true benefit from an effective treatment; or one trial might falsely find a benefit from an ineffective treatment, either by chance, or because the study was designed so badly that it not longer represented a “fair test” of the intervention, against whatever you were comparing it to.

The Cochrane Library is an international non-profit collaboration of academics that brings together all the evidence on a given question, using a predetermined standard method for seeking out information, assessing its quality, and combining it into one giant report. They’re slightly turgid, and they are considered by medics and academics to be pretty much the best quality evidence available, because they look at all the relevant trial evidence on a given question, although you will very rarely hear about Cochrane reviews in mainstream media, because journalists are seduced by baubles and novelty.

Handily, there is a Cochrane review on Tamiflu, and a similar drug called Relenza. In reality the drugs’ names are oseltamivir and zanamivir, but for some reason the media always use the original manufacturers’ brand names instead of the generic, a bit like calling all ibuprofen tablets “nurofen”, or all aspirin tablets “dispirin”. After a few years all medicines come out of copyright patent, at which point anyone can manufacture them, but if everyone is used to the brand name rather than the generic then the original company has an advantage. Anyway.

The review on oseltamivir and zanamivir was done several years ago, but reviews are frequently updated in the Cochrane Library, because evidence changes. This review was properly re-done in 2006, and also checked as being up-to-date in May 2008. They asked two questions. Do these drugs treat flu? And do they prevent it?

The time to alleviation of flu symptoms was assessed by nine trials, and the pooled hazard ratio for zanamivir was 1.24. Sorry, that was gibberish: what I meant to say is that the group treated with zanamivir are 24% more likely to have their symptoms alleviated than the placebo group, at a given time-point. For oseltamivir the figure was 20%. It’s alright. I’d take it. It’s just not amazing.

The NICE review from February 2009 looks at similar data, and analyses it in a different way, giving you absolute time to recovery, which is a little easier to understand. Overall, oseltamivir reduced the average (“median”) time to alleviation of symptoms by 0.68 days. For zanamivir the figure was 0.71 days. So you get better 16 or 17 hours sooner if you take these drugs.

The prevention studies are a bit more exciting. Patients had less virus on board, and less in their noses, but neither drug actually stopped patients being infectious. In fact, neither drug had a protective effect at all against influenza-like-illness, or asymptomatic influenza, even at higher doses. (“Influenza-like-illness” is symptoms that look like flu but might be a bad cold: it can take days to get blood tests back, so you treat it as flu, and in a pandemic, you can be more certain that you’re seeing real flu, because there’s more of it about).

For preventing symptomatic influenza, the results are more impressive. Oseltamivir 75mg daily was 61% effective compared with placebo, and 73% effective at 150mg daily, while Relenza was 62% effective. In trials looking at preventing influenza in people who were living in households where someone was already infected, the drugs were also pretty good.

I’d take these drugs if I needed to. Things might be different in a pandemic, and the Cochrane review recommends them in this situation. If they make my symptoms less severe then I’m guessing I’m less likely to die, for example, and they might reduce the spread through a country.

But they’re not miracle cures, and if this is worrying to you, then that just shows how ill-equipped we are to think about risk. For the question of whether we’ll see a pandemic, things are so up in the air that it’s not even possible to give a number defining that risk (which many people seem to find incomprehensible). For the efficacy of the treatments, we do have numerical risk data on the chances of getting better, but for all interventions we have to accept that modern medicine is all about shaving risks and probabilities, in lots of different ways, to achieve the best possible outcome, rather than absolute certainty. And after all that, it’s true, if you got swine flu, you might still die. Which would be seriously rubbish.

Neuraminidase inhibitors for preventing and treating influenza in healthy adults. Jefferson T et al. Cochrane Database of Systematic Reviews 2006, Issue 3. Art. No.: CD001265.

NICE Technology Appraisal Guidance No 168 – Amantadine, oseltamivir and zanamivir for the treatment of influenza.