Data sharing

Posted by Jean Adams

As you may know, I am fairly relaxed about the idea of sharing data about me for the purposes of research.  But now I find myself in a dilemma about sharing my research data.

The vast majority of public health research funding in the UK comes directly, or indirectly, from government, or charities. As someone who has benefitted personally (they pay my salary) and professionally (they pay for my research) from this public funding, I feel some duty to remember who is paying for it all. I make efforts to share my findings with the public, and have tried to learn how to do this better. I believe strongly in the very concept of Fuse and the need to engage with those working in public health policy and practice to ensure that my research is relevant and usable, and is used.

The research data produced as a result of my public funding is not my data, it is our data. I know this. And yet now that I have been asked to share some, I keep trying to think of reasons not to.

Sharing is caring

A few years ago, Ofcom introduced new regulations on the content and scheduling of television food advertising to children. A complex definition of what constituted ‘unhealthy’ food was devised and the regulations prevented advertisements for these foods during programmes that have particularly high child viewerships. We were awarded funding from a consortium of public and charity funders to explore the impact of these regulations on what advertisements children were seeing on TV.

The research was very data intensive. We bought a large spreadsheet of information describing every single advertisement on UK television over four different weeks – before, during, and after the regulations were implemented. From among these 2.5 million adverts, we picked out the food ads and collected separate information on the nutritional content of every food advertised. Next, we worked out which foods met the definition of ‘unhealthy’ and compared the number of ads seen for unhealthy foods across our four study weeks. You can imagine that it needed a very special sort of person to have the patience and attention to detail to do all of this. It took her a year.

Our findings were published last week and there was a small amount of media coverage. I thought it wasn’t enough coverage, but it was enough to attract the attention of a snack food marketing organisation. Now they would like to know if our data could be used to track changes in snack food advertising. They are even suggesting that they could pay for a researcher to perform such an analysis.

I feel deeply unsure about getting involved in this. As someone who has real concerns about the public health impacts of the food industry in general, merely speaking to these people feels like dancing with the devil. But, presumably my publically funded data belongs to British industry as much as it does to the British public.

Right now I am employing delaying tactics – and not just writing this. Soon I will have to make a real decision about what to do.

On air...again

Posted by Jean Adams

Yesterday I got up at 0530 to speak to someone from Radio 5 Live for their breakfast programme. Then I made and ate my porridge and drove to the airport for the early flight to Bristol. I am so not a morning person. But my morning, day, week, month, year was totally made by hearing my research mentioned on the Today programme during the 0630 headlines. I nearly crashed the car.

 
About 18 months ago, I was asked for ideas for final year projects for Food & Human Nutrition students. I don’t remember now why I thought it would be interesting to look at price promotions on alcohol. But I did. I was given two students to do the project. Between them, they surveyed all the shops selling alcohol within 1500m of Newcastle’s Student Union, noting down all price promotions they saw on their rounds. The students split up the data and wrote up their dissertations.

When I put the all the data back together, it became clear that of more than 2000 promotions found, less than 2% led to alcohol being sold at less than cost price – the new minimum price that will be introduced in England in April. The implication is obvious – banning below cost sales of alcohol wont effect the cost of alcohol in the shops. The research was accepted for publication in Alcohol & Alcoholism.

I told the press office about the research because it felt policy relevant to me. I’m not sure they were desperately interested, but they agreed to press release it.

I ended up doing TV interviews for BBC Breakfast, Tyne Tees and Sky Tyne & Wear. As well as Radio 5, I did Sky Radio too. The research was mentioned without interviews on Radio 3 and 4 morning news. But by far the most adrenalin-surging was a live telephone interview with Winifred Robinson on Radio 4’s You & Yours.

It feels so unfair that the media loved my quickly done, undergraduate students projects paper, and dismissed my really important, publically funded evaluation of new regulations on TV food advertising that took years to produce.

Well...I am in Bristol for the third instalment (and final exam) of a Science Writing course. So I think this all has something to do with ‘news values’. Our work on alcohol promotions plays to the “continuity” and “consonance” news values – it builds on David Cameron’s statement last week that he was going to tackle the “scandal” of binge drinking, but it also allows well rehearsed arguments about minimum unit pricing to be dredged up again. All in all, it’s a nice little piece that journalists can easily relate to other ongoing public debates. Can you tell I’ve been revising?

So now I can diagnose the problem. But I still don’t know how to treat it and get the media to take note of stuff they just don’t want to.

On air

Posted by Jean Adams

I was on BBC Radio Newcastle this morning with Charlie and Alfie. We were talking about some new research that we’ve just published on the effects of regulations restricting TV junk food advertising to kids. It was kind of fun. Charlie and Alfie seem to just mess around a bit, link to the weather and traffic, and sometimes talk in a light hearted way about semi-serious subjects (is it bad that I’ve never listened to the local radio breakfast show before?).

Charlie & Alfie

Our research explored the nutritional content of food being advertised on UK TV before and after new regulations came in restricting advertising of less healthy foods during children’s programming. We found no difference in the amount of advertising for less healthy foods that kids were seeing following the regulations. But the regulations were widely adhered to. The explanation? Kids watch more than just children’s programmes.

The work has also been covered by BBC News Online, Sky Tyne and Wear, and some other places you’ll likely never have heard of.

I think the work is really, really important: regulations on TV food advertising are adhered to by broadcasters, but as they stand they have no effect on kids exposure to unhealthy food ads. Implication: extend the regulations to all TV and they will work. The study also draws attention to an, ineffective, government initiative that both Labour and the Coalition have been quite enthusiastic about.

I spent quite a lot of time working with the University Press Office on a press release. The press officers tried pretty hard to get a variety of journalists interested yesterday. Local radio and BBC online is so not good coverage. So why doesn’t anyone care about my baby research?

I haven’t quite worked out the answer to this yet. But I think there might be a variety of reasons. Firstly, the whole idea that adverts and marketing might alter our behaviour is pretty difficult – it rather challenges the idea of free-will. Next, even if food adverts do effect what our kids eat (they do), it’s probably only a small effect amongst many other (also small) effects. So it’s a difficult topic to get excited about. Also, I’m wondering if the ‘no effect’ message is just not that exciting – ‘kids seeing more junk food ads, despite regulations’ might have been a much more interesting line.

But what I’m starting to think more about, is that the problem might be related to how the media works. Our university press officers primarily work with biomedical scientists. They are used to targeting science journalists with releases. Is an analysis of the effect of food advertising rules science? Or is it maybe consumer studies, or some other sort of wishy-washy social science? Personally, I don’t really care what it is, and I’m happy to be a multi-disciplinary public health researcher. But if journalists at national daily outlets only cover their own beat, and there’s no beat for multi-disciplinary public health research/wishy-washy social science, I guess we’re screwed.

Peer review

Posted by Jean Adams

“Dear Dr Adams - Given your extensive expertise in the area, we would be grateful if you would be able to review this paper.” 

Peer review by Josh

Another day, another flattering invitation to peer-review a manuscript submitted to a journal. Sometimes they even call me Prof. Adams. Time was when I thought that it was pretty cool to be mistaken for Prof. Adams. These days I just think they were probably too lazy to check.

Peer-review is self-regulation for scientists. You do your research. You write it up and send it to a journal. The journal editor asks someone else an independent expert to check that the research and write-up are kosher. Depending on the peer-review report, your manuscript is either accepted, returned to you for corrections, or rejected out of hand.

The idea of peer-review is that mistakes will be identified, flaws in an argument spotted, and over-enthusiastic interpretation of results calmed before a manuscript makes it to print. But like self-regulation, peer-reviewed is notoriously flawed. Not only do genuine mistakes make it past peer-review, but so too do serious errors with important and ongoing implications.

Despite this, peer-review is an academic’s life blood. Unless my papers are peer-reviewed, they will not get published. So I feel an obligation to contribute to the peer-review roundabout and to review papers that are sent for my ‘expert’ assessment. Even when the abstract sent with the invitation to review looks like the paper is likely to be nonsense, I routinely accept: because I submitted a paper this week too, and because if I don’t, who will?

So what am I to do when a 5000 word paper promises to tell me about the relationship between variable x and “human behaviour”? Should I send the paper back with a brief note pointing out that no 5000 word paper could possible ever hope to explore the relationship between anything and all human behaviour? Or should I painstakingly go through the paper and identify each over-simplification, and each wild assumption, line-by-line, point-by-point?

One of the problems with peer-review is that you are rarely given any instructions. I genuinely don’t know what I am supposed to do in this situation. Mostly I am super-conscientious and end up doing the point-by-point response, but at the same time feeling like I have been duped into providing supervisory feedback without having agreed to supervise.

I recently started following @richardhorton1, the editor of the Lancet. He is full of hilariously bombastic wisdom:

“The mistake, of course, is to have thought that peer review was any more than a crude means of discovering the acceptability—not the validity—of a new finding. Editors and scientists alike insist on the pivotal importance of peer review. We portray peer review to the public as a quasi-sacred process that helps to make science our most objective truth teller. But we know that the system of peer review is biased, unjust, unaccountable, incomplete, often insulting, usually ignorant, occasionally foolish, and frequently wrong.”

So maybe I should just stop doing peer-review at all? Then what would happen?

The REF

Posted by Jean Adams

Sometimes when you work at a university you get lured into thinking that the REF is the only thing that matters in the whole world.

REF = research excellence framework: “the new system for assessing the quality of research in UK higher education institutions”.

Ref gives Pablo Zabalete red card 

Once every 5 years or so the whole UK university system evaluates itself. Each department at each university submits information on its work since the last assessment exercise. Panels of experts make judgements on the scientific importance of all the work submitted and every department is ranked for excellence within its field. Then the central university funding agency uses the rankings to decide how much money each university gets. 

When I say they review our ‘work’, I primarily mean they review the papers we’ve published in scientific journals. Some other things are taken into account, but mostly it’s our journal papers that matter. Each academic submits four papers and the expert panel rates them as world leading, internationally excellent, internationally recognised, nationally recognised, or zero. There is no difference here between doing work that is not ‘nationally recognised’ and not having bothered at all.

Cue academics getting their knickers in a big twist about what ‘world leading’ means and how we can demonstrate that our publications are ‘world leading’. Not so long ago, we got totally obsessed by the impact factor. This is a rating given to every scientific journal that is simply the average number of times a paper published in the journal is referenced in another paper in a scientific journal. 

The assumption was that if your research was published in the sort of journal that people tend to reference a lot then it was likely to be pretty good research. Yes really. That really was how we decided if research was good or not. Not anything like will it save a life, or make the world a better place, or improve out understand of human nature. Just: is it published in the sort of journal that other people mention in their work? There are even rumours of journals explicitly manipulating the system by only publishing papers that referenced other papers in the same journal – presumably specifically to help increase the impact factor of the journal.

But for REF2014 we have a totally new and improved method of deciding how good your work is: the citation count. It ignores anything to do with the journal itself and is a simple count of the number of times someone else references your paper in another journal paper. Yes really (again). Even after thinking about it quite hard, and recognising that we might have got it wrong last time, this is the best a bunch of clever people can do for determining how well we’re all doing.

In my world, the only thing that matters is if someone else is writing about me. You are welcome to write: “Adams et al., (2009) were clearly talking nonsense”. But please do remember to write it.

Research governance - or is Jean a geeky weirdo?

Posted by Jean Adams

It took me a while to work out what ‘governance’ means – and a lot of “I know this might be a stupid question, but…” My conclusion is that research governance is about maintaining high standards of research conduct. This includes not breaking the law. And who would have known there were so many laws that I could break, sitting here in my university cubicle, geeking around with data and statistics?

This week’s saga involves linking disease register data to Hospital Episode Statistics (HES) data. Now stick with me here, because I hope this is going to be more interesting than you think right now.

A disease register collects information on all the people with a particular disease in a particular area. Fuse has access to quite a few different disease registers. Hospital Episode Statistics is a massive database of information recording every treatment delivered in hospitals in the UK, every outpatient appointment that takes place, and the diagnoses of every patient involved. It’s geek-heaven.

Medical records

-->
So, we would like to link our disease register to HES to work out what other illnesses people on the register suffer from. This is possible because both the register and HES use NHS numbers to identify individual patients. We know the NHS numbers of the people of the register, and we can use these to look up the same people on HES.

The problem is that we do not have patient consent to link the registry data to the HES data. To be honest, I bet you no-one has ever asked your consent to collect the HES data. And, to tell the absolute truth, no-one asked consent of the people on the disease register to include them on it. This is because all of the information in both the disease register and HES is extracted from routine medical records. Your doctor writes in your notes that you have a particular disease and so your information is sent to the disease register. You get admitted to hospital with breathlessness and someone writes in your notes that you had a chest x-ray and were treated with antibiotics and this information is sent to HES.

This turns out to be legally challenging because us geeks are now asking to use that routine medical information for research. Your information was collected by medical professionals for their work and to maximise yoru care. That's generally accepted to be okay. We now want to use it for research and the legal position is that that's not okay unless we ask you first - or seek special permission for the law to be waived on this occasion.

There is one more thing I need to tell you here. I, and none of my colleagues, will ever be able to tell who the individuals on the disease register are. The hospital that holds the register stripped out all the names, addresses, and dates of birth before they let us have it. Similarly, if we ever get hold of the HES data there is no way we will ever be given names, addresses, dates of birth or anything else that would allow to tell who was who. The data provided to us is entirely anonymous.

Probably quite a lot of people have no idea there was a massive UK hospital data warehouse or that disease registers exist. But now that I’ve told you, you might not be that surprised? What I’m wondering is: are you bothered? Do you care that there is information on all your hospital visits? Do you care that someone could link it to something like a local disease register to do some research? Remember, it’s all totally anonymous. All it is is a spreadsheet showing that person number 73 is male, attended an cardiology outpatient appointment in September 1989, was started on blood pressure lowering medication in October 1989, was admitted with chest pain in July 1993, blah, blah, blah.

I don’t care that much. But that’s maybe because I’m a geeky weirdo.