When interviewing Dr Goldacre, the poor chap was enduring gastroenteritis so like many, including yours truly, he persevered with work whilst suffering the effects. The experience was like watching Billy Connolly doing live stand-up. Like Connolly, with his fiercely intelligent and keenly perceptive mind, Goldacre will wax lyrical about something that he is passionate about, then wander off on a tangent, and then return to his subject matter with a clear cut conclusion. (And they both have outstanding hair).

So here Dr Goldacre talks on everything from Bad Science to AllTrials to Twitter to the role statistics has to play in medicine. Sit back and enjoy. I did. And sign the AllTrials petition.

1. You have a very active Twitter feed. What do you like about Twitter and how important has social media been in giving Bad Science a voice to raise awareness of issues? Are there any key successes you can highlight?

What I love about Twitter is that you can follow a thousand people who are all working in different and interesting fields and just every now and then, trail your fingers into this river of information as it passes by. Also, as it is so low threshold, you gain access to the passing thoughts of very clever people which you never previously would have had. In the past, if Richard Horton, Editor of The Lancet, was waiting in the check-in queue at Heathrow and was reading a global health story in a newspaper which he thinks is wrong, that thought would have eventually disappeared. But now, he can pull out his phone and tweet that the news story is incorrect as the The Lancet published an article on that subject some months back.

So being able to sift through those tailored feeds of serendipity is absolutely amazing and I think it is also really interesting for showing you who is truly clued up on their subject matter. I have a deep-rooted prejudice which is that if people can talk fluently in everyday language about their job, it strongly suggests that they have fully incorporated their work into their character. They feel it in their belly. There are people with whom you talk about technical stuff and it almost feels like they can only talk about it in a very formal way with their best work face on – as if the information they are talking about has not penetrated within. Twitter cuts through that and is a way of finding people who are insightful and passionate about what they do, like junior doctors one year out of medical school who take you aback when you realise they know more than people whose job it is to know about a particular field, such as 15 year-old Rhys Morgan. He has Crohn’s disease and went onto Crohn’s disease discussion forums and discussed evidence, whilst noting down people making false claims about evidence for proprietary treatments. He ended up giving better critical appraisal of the evidence that was presented than plenty of medical students. This was all simply because he read How to Read a Paper by Trish Greenhalgh and some of my writings, so he has learnt about how critical appraisal works and what trials look like along with the strengths and weaknesses of different kinds of evidence. Thanks to Twitter, I have been able to read about people like Rhys in action and to see ideas and principles really come alive and be discussed and for that, it is wonderful.

There are people out there who want the hardcore campaign on trials transparency, others who want the laurels about Gillian McKeith. You realise that you are a bundle in the same way that newspapers are in that they can spend so much money on showbiz and gossip but in fact, the way that newspapers get people through the door is via this very gossip and they then divert the reader’s attention to think about more substantial issues. Twitter can be annoying, like when someone you follow tweets endlessly about cricket but that’s life.

2. AllTrials is a campaign which calls all past and present clinical trials to be registered and their full methods and summary results reported. How did this project come about? Are you able to measure its progress so far?

AllTrials is not really about open research in the sense of open lab books or anything technical. Its message is this – if you have completed a clinical trial, you have to share the full methods and summary of results with doctors, researchers and patients because that is the only way we can make truly informed decisions about which treatment works best.

It was precipitated by claims from people in the industry around 2012 that the problem of trials going missing in action no longer existed and that it had been fixed. This was in response to my book Bad Pharma. I was kicking around on my own trying to get people working in government to take it more seriously and simultaneously, many leading names in academia had been trying to raise these issues for some time, such as Iain Chalmers and various people at the BMJ. It seemed to me that we needed to work in a constructive and collective way in order to fix this problem.

There have been 70-80 prevalent studies looking at the number of completed trials that do not report their findings. John Snow found that cholera was caused by excrement in water but he didn’t then publish that and stop and award himself a point. He took the handle off a Broad Street pump and triggered a change in global sewage policy and that is exactly what we need to do with the issue of trial registration and results dissemination. Getting that kind of structural change in medicine is a professional job. If you talk to people who have trained in public health, they are often taught in a boring, tedious way about how you identify a problem in data, then identify key stakeholders, find where the opportunities and barriers are, and you set up meetings about how to get things done. In the area of publication bias, to me, we had been expecting that just publishing papers would fix the problem when it clearly hasn’t.

I am a real nerd for information architecture and structural issues in evidence-based medicine and for far less than the cost of ONE trial, we could fix the information architecture issues for ALL trials.

So we hooked up with Sense About Science, a professional lobbying organization and a really great interesting charity who previously mocked quacks in science but also worked with others in helping to take the public engagement with science more seriously. They also worked on the libel reform campaign which came about after my friend, Simon Singh got sued for libel over some unkind things he said about chiropractors and that was a couple of years after I got sued by a vitamin pills salesman called Matthias Rath in a case which was unsuccessful but it cost the Guardian £500,000 to defend. We got a change in the law and it was clear to me that Sense About Science was strategic and cut straight to the heart of the matter and that’s what we needed with clinical trials.

In terms of our impact, one can quickly measure process outcomes e.g. we can count the number of meetings we have had with MEPS, journalists, etc.; we can say for definite that we have triggered various stories; we have seen our policy analysis being picked up and used by others. We are just about to launch our trials transparency index which is an audit of the 50 largest global drug companies. To be clear, drug companies are not the only bad people here because non-industry sponsored trials are often appalling at sharing their results too but what is interesting about drug companies is that they are the only people who actively campaign against transparency. This makes them out as the enemy which I don’t think they are – there is a range of views, policies and activities within the industry and what we have done with the trials transparency index is set about trying to map that.

As with any audit cycle, we set out the gold standard – all trials registered, how far back do you go, do you have independent audit of compliance, the same with study reports, clinical reports, patient data – then we have applied a scoring system whereby you obtain marks for how much you do and how far back you go. In turn a scoring system is applied to this as well which produces a ranking just like the Access to Medicines index. What is lovely about the process of doing this are 1) everything we ask companies to do, is done by at least one company, 2) there is a huge range so some companies are fantastic, others are terrible. We have to remember this is based on policies, not actual performance. As with all audits, I think you have to assume that there will be good faith from some players even if not all. One really good aspect about the audit is that people who are performing badly can learn from people who are performing well. This is one of the best aspects about GlaxoSmithKline, the biggest drug company in the UK, who have signed up to AllTrials and they have really knocked the legs out from underneath the spurious objections of making all results available, putting a stop to arguments such as it being too expensive to do this, when it is not.

We have also been working with ethical investors and we have been successful in engaging with them via a couple of seminars. They are excited to see our first trials transparency index and we will be updating it annually.

The combination of Bad Pharma and AllTrials has made it absolutely impossible for people to pretend that this problem does not exist and also by sheer force of reputation, we have made it impossible for people to hold various other positions on it. I think that much is down to us. We will see what happens next. It seems amazing to me that we have allowed this situation to exist and reflects a real cultural blind spot because we would recognise if you deleted half the data points from one study, that would be misconduct but when whole studies are withheld from decision-makers, that has exactly the same impact on the apparent benefit of the treatment and yet, people have not seen this as being as important.

It makes no sense in terms of efficiency. We do randomised trials because we want to exclude bias and achieve the most accurate estimate of the benefits of treatments that are being compared. We spend so much money (on average $10K per participant) to exclude this bias and then at the crucial moment of adding all the results together in order to obtain the best overall answer, we allow a third of trial results to be withheld from doctors. I am a real nerd for information architecture and structural issues in evidence-based medicine and for far less than the cost of ONE trial, we could fix the information architecture issues for ALL trials.

The fact that we have not done so and the fact that there is so much resistance to do so is culturally interesting as senior members in academia and medicine really should know better.

We are just about to launch our trials transparency index which is an audit of the 50 largest global drug companies. To be clear, drug companies are not the only bad people here because non-industry sponsored trials are often appalling at sharing their results too but what is interesting about drug companies is that they are the only people who actively campaign against transparency. This makes them out as the enemy which I don’t think they are – there is a range of views, policies and activities within the industry and what we have done with the trials transparency index is set about trying to map that.

3. You have won the Royal Statistical Society’s award for Statistical Excellence in Journalism Award. Do you consider yourself a journalist and of all your research, which was the one where you were most taken aback by in terms of the misuse of statistics?

I don’t think I am a journalist. I sort of resent those kinds of labels generally. In medicine, we are very careful to not say ‘a schizophrenic’, we say, ‘a person with schizophrenia’ and equally, I feel anxious about nouns denoting everything about somebody. I am also aware that the way that I am introduced is completely random and depends on where you are and whether someone likes you or not. If they want to criticise me, I’ll be called a blogger and activist, or a psychiatrist is the worst people can manage, GP even worse! Even when people want to praise you, they’ll call me by some formal title I am not sure I like the sound of.

The worst misuse of stats ever? There are so many. What helps me decide whether something is worth writing about is not just whether it is wrong, is it interestingly wrong? Is it wrong in the sense that it illustrates an important structural problem or methodological issue? I don’t care if the UK government spends half a billion pounds on Tamiflu; I do care whether they had access to all the results of all the trials when they were making that decision. I really care that everybody also had access to that information so that we could critically appraise that evidence and decide whether we agree with the UK government’s decision. I fully recognise that human decisions in health are often completely arbitrary and driven by fear and politics as much as they’re driven by evidence. What matters about Tamiflu is that it is an illustration of a broader structural issue – a lack of transparency but also a lack of structured approach to transparency for clinical trials and all decision-making.

4. What kind of feedback do you receive from your audience? In terms of positive feedback, do they include statisticians or members of the public who applaud you for spotting these inconsistencies? How about the negative?

Yes, people like what I do and there are also people who despise me! You have to be really careful if you have any kind of public profile because you need to keep listening to feedback and yet ensure that the positive and negative stuff doesn’t really hit you. I think that is one nice thing about being a doctor and working in difficult areas for a long time, in that you develop a degree of equanimity. There is a really good interview with Martin Amis who said that when someone is firing bullets at you, you have to wear a bulletproof vest but just before the bullet hits, you have to grab it and check what is written on it. The approval of strangers is clearly intoxicating and you see people being taken in by it. You have to try not to get caught up in it and when people say horrible things, it is pretty easy to ignore. If someone is cross with me on Twitter, I sometimes check their profile and it turns out that that person is cross with everyone. You also have to read and make sure that someone has not identified something where you have been either flat wrong or more commonly, that there may have been nuances or other more interesting contextual issues that you had missed.

My email address is ben@badscience.net and it is in the back of each book and at the bottom of every article I have ever written. My email inbox is just a joy every day because it is people sending me something that I may be interested in from…”Do you want be involved in this?…I am trying to get this fixed…Do you have any advice on how to go about it?…You’ve got this wrong”. It’s great but I have to be careful not to be dependent on it. Your family, friends and colleagues matter at the end of the day and it is not hard to cut yourself off from childishness on the internet.

5. Last year, I interviewed Sir David Cox who commented – “I would like to think of myself as a scientist, who happens largely to specialise in the use of statistics”. Do you agree that statistics is less of a discipline in its own right, and more of an integral part of science as a whole?

Functionally, in my experience, statisticians are like a cog in a bigger system. It’s a great place to be actually. The best statisticians that you work with are good at explaining things to people who are not so good at grasping the issues. They understand what a plausible explanation looks like, and are capable of identifying and following good advice.

I would hope that statistics is becoming an integral part of science. When you look at how statistical tools are routinely misused, it would be great if it was more thoroughly embedded. As a discipline, the fact that it is a cog in a bigger system and has an air of service about is very healthy, in my opinion. I view my work as being a service – I want to make practical tools that fix problems and I want communicate to people to help them understand better. I think one of the biggest problems with academia is that it can drift into being sometimes too self-serving and one of the healthiest aspects of medical academia, in particular, is that, at its best, it is pragmatic and patient-oriented. If statisticians are part of a bigger web of people trying to find the best answers to important questions that has to be a good thing. Obviously you still need stand-alone theoreticians to come up with ways of doing regression analysis etc. that I feel that I almost understand after an hour of reading a paper (if I’m lucky!). The bulk of the work is service and in academia, we ought to be more service-oriented and more thoughtful of what we can do that would be of immediate practical benefit.#

Without question, the most exciting thing to happen to statistics in the past ten years is the democratization of access to tools for analysis and raw data.

6. What has been the most exciting development that you have seen in statistics during your career so far? (Have there been, in your view, any exciting developments in statistics?)

Without question, the most exciting thing to happen to statistics in the past ten years is the democratization of access to tools for analysis and raw data. I think that brings enormous opportunities but also enormous challenges, so the fact that anyone can download a copy of R, download a bunch of data, and then find patterns is truly amazing and has given us some extraordinary things. It also produces the opportunity of creating spurious findings, in particular, publication bias. It really breaks assumptions that you make when you use statistical tools but at the same time, it is beautiful and extraordinary that people do that.

I find it really fascinating that school-level teaching on statistics has completed failed to catch up. What you want to be teaching to schools about statistics is ‘Here is a big messy data set. Let’s talk about how you are going to approach it and find some useful patterns that might actually benefit somebody. If there are shortcomings in what you are able to do with it, then tell me how you might improve the data collection in order to make it better.” We need to teach that kind of practical statistics, which I think is often neglected, but also at the same time crucially demonstrate that statistics is about data analysis at a scale that goes beyond an equation on a piece of paper. Our failure to do that is really tragic. What use is half the stuff you learn in school level maths? Having a better understanding of how to extract information from dirty data sets and spot patterns, combined with a better understanding of where the pitfalls are and how people might mislead you is something you can definitely use in later life. Also it would really benefit business, society, everyone everywhere if there were more people around saying “You know what? I am going to apply the basics of statistical process control to my corner shop/school/hobby”, whatever. There are real opportunities for people who might have a use for those kinds of tools and to deprive people of those skills and insights on a national scale is criminal.

7. What do you see as the greatest challenges facing the professions of medicine and statistics in the coming years?

I think in medicine and statistics it is really obvious. We need to up our game and obtain structured, shared, interlinked data on everything that we do. You can see that people are making baby steps towards that e.g. ORCID IDs for authors so that you can match up papers and the standard results from clinical trials. You can imagine a vast interlocking network of data from evidence-based medicine that would support shared decision-making in a clinical setting. This would then free up doctors to do what only people can do with patient interaction – to talk about priorities and benefits vs side effects, etc. I think the real challenge for doctors and statisticians now is how we can create an information architecture of the process of gathering and disseminating knowledge to doctors and patients, so that they can then have a chat on the best techniques. That is something that we are insufficiently ambitious about in medicine and that is partly what I am doing here at Oxford. There are hundreds of people around the world working on these issues but they don’t get the same amount of money as one individual clinical trial and that is very peculiar. That is probably our fault because we don’t have a better catchphrase than the information architecture of evidence-based medicine.

8. Are there people that have been influential in your career?

Family – I am a fourth generation nerd. In fact, my dad is an epidemiologist and my mum is a pop singer. So basically I have become a stage epidemiologist – I have literally taken one word from each of my parents’ job title and put them together. Iain Chalmers, who founded the Cochrane Collaboration who is one of the pioneers of evidence based medicine. Probably my greatest influence is the giant cloud of people out there on Twitter and email, and anyone who has ever asked a question at the end of a talk. In reality, I feel very fortunate to have been able to become an idealist intellectual at this particular moment in history. When people say that you are part of a great conversation, it sounds naff but being able to throw ideas out there and receive immediate feedback, such as “I didn’t quite get that, could you explain that better?” to “Wow, that is really interesting” for something that I had side-lined at the bottom of a piece – it is just amazing. I couldn’t be happier so realistically my greatest influence is the entire world of amazingly sharp people in all professions and roles that fly into my life through that laptop and that phone. God only knows what comes next but this is amazing.