Evidence based voting

April 24th, 2010 by Ben Goldacre in bad science, politics, structured data | 53 Comments »

Ben Goldacre, The Guardian, Saturday 24 April 2010

What can science and evidence bring to an election? First there are the micro-issues: we can assess the validity of claims made by politicians by seeking out the evidence. David Cameron, for example, claimed that UK cancer services were bad because fewer people die of cancer in Bulgaria than in the UK, which many have already debunked: he used death data from a country with inferior monitoring standards, and a far lower life expectancy, but more than that, he used death data, which is driven not just by treatment success, but also by the number of new cases to start with, which can vary widely for all kinds of reasons.

Still on cancer, but in a neglected example, Cameron also said: “I have a man in my constituency [with] kidney cancer who came to see me with seven others. Tragically, two of them have died because they couldn’t get the drug Sutent that they wanted.” This was odd, because Sutent is actually available on the NHS for kidney cancer, having been approved by NICE in February 2009 (despite costing £54,366 per “quality adjusted life year”, while NICE generally draws the line at £30,000).

Where did Cameron get the idea it was rejected? I don’t know. On the 9th of April the Daily Mail published a bizarre article claiming that 15 cancer drugs (including Sutent) had been rejected by NICE. They provided a list. Bizarrely, 10 of the 15 drugs on that list which the Mail say NICE rejected have actually been approved by NICE. The head of NICE wrote to all three parties on the 9th of April explaining that this article was bizarrely wrong. His letter is posted in full on the NICE website. (NICE did reject Sutent for people with a poor prognosis who are unsuitable for immunotherapy, but only there because they weren’t given any evidence on its use in that group).

As a side issue, of course, it’s always unwise to claim that someone would be alive today if they had a specific treatment, in any situation. The mighty statistician Prof David Spiegelhalter has run through the maths on his Understanding Uncertainty website, and even if you are in the better prognosis group, for whom we do have evidence, you only have a 58% chance of surviving longer after receiving Sutent, and the median survival is 37 months with the drug and 27 months without it. Worth it, of course. I’m just saying.

Alongside the science of individual claims, it’s also worth looking at what the parties say about science itself. Here, only the Liberal Democrats’ manifesto stands out, promising funding like the others, but also making pledges on the use of independent scientific advice in policy (in contrast to the David Nutt affair), and even coming out in favour of open-access academic publication.

But more than anything, this election offers a new opportunity: beyond expecting our politicians to follow the evidence, we can be evidence based ourselves, in our voting decisions, and hold ourselves to the same high standards. VoteMatch can match your views against the party manifestoes. SkepticalVoter knows if your MP believes in the homeopaths’ magic beans. Using TheyWorkForYou I can see that my last MP (Andrew Smith, Labour, Oxford East) voted very strongly in favour of the Iraq war, very strongly against an inquiry into it, very strongly in favour of ID cards, and so on. It’s grim reading. PoliticsPosters will print that on a poster, so you can make sure your neighbours know too.

Websites like these recognise that disseminating data in a meaningful form is as important as gathering it in the first place. Your MPs voting record, and their attendance in parliament, have always been publicly available, but until recently it was an effort to find, so only journalists and lobbyists knew the ugly details. That has changed. In the past we relied on hunches and reputation, but now – uniquely – we have the option to leave tribalism behind, and deploy our vote using data.

If you like what I do, and you want me to do more, you can: buy my books Bad Science and Bad Pharma, give them to your friends, put them on your reading list, employ me to do a talk, or tweet this article to your friends. Thanks! ++++++++++++++++++++++++++++++++++++++++++

53 Responses

  1. maninalift said,

    May 5, 2010 at 10:02 am

    On the subject of politically potent but far from proven arguments is the “big, centralised, bureaucratic organisation is inefficient” argument which DC keeps banging on.

    Large organisations inevitably have pockets of inefficiency and it is necessary to have different structures across an organisation for self-examination to eliminate these, however, in my experience by far the greatest source of inefficiency in the NHS comes from the market structure.

    Indeed the more autonomy one gives to unit parts, the more means it is necessary it is to have to direct those parts and the more conflicting forces there are on it’s staff: It is not at all clear that it leads to freedom for medical judgement or from paperwork.

  2. maninalift said,

    May 5, 2010 at 10:03 am

    sorry for the repeated posting – I blame wordpress.

  3. three tigers said,

    May 6, 2010 at 7:20 pm

    Remember here in Switzerland 20.9% of the population are foreigners, in a total population of ~7.2 million. My ex-pat family included, although our kids were born here. The SVP (or Swiss Nazi Party – as I like to call them) did a campaign which would basically not be allowed in the UK for its obvious racism to scaremonger about the minaret initiative. A Swiss lab technician of mine jokingly said he would be happy to vote for it, but only if they include all the bloody churches too and their bells! Many Swiss people are not so rich, or so homogeneous as outsiders think, but many are very conservative, rural farming folk and natural supporters of the SVP (a bit like farmers in the UK being Tories). However, having observed both (and if our Swiss citizenship is approved) being able to vote in both, my vote here would potentially count for more.