Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
When is a Picket not a Picket? It's not a joke – or at least, I can't think of a very good punchline if it is. You will recall that Picket is our section which looks for sentinel papers, re-writes the abstract in a structured way, and then puts this paper in the context of the broader literature with a commentary. It is an approach which we ripped off from the earlier layout of the journal, Evidence Based Medicine, and which we – and our feedback from readers would confirm this – think works fairly well. We have written elsewhere about how we go about selecting papers to abstract1 but of course sometimes we hear of papers in other ways. Sometimes an article has made quite a splash in the media, and this is where we hit a bit of a problem. These are important and interesting papers, because our patients and their families will also be hearing about them. But as I am sure you will know, they are not always the best.
This issue includes a Picket on one of the weakest papers we have yet to cover (see page 78). Actually, that is not very kind; it is probably perfectly sound as a paper, just hugely over-exaggerated in the press. It provoked quite a debate among us about when a paper should not actually get Picketed; we learnt a lot from the error we made here, and we would not Picket this paper if given it today. Almost exactly at the time we were launching the Picket series, in August 2010, there was a news story carried in much of the mainstream UK media, based principally, and fairly uncritically, on a press release from Kings College London and the Medical Research Council, which you can find here: http://bit.ly/AutismPRMRC. In short, the release described that a brain scan could diagnose autism with ‘90% accuracy’, and the radio interviews speculated that things would be even better in children. I am sure I was not alone in trying to dig out the paper, sure that it would only be a fairly short time before I had a family request the scan from me. Of course, I was also pretty interested to understand what ‘accuracy’ meant, assuming I had missed that lecture at medical school.
The true story is quite a different – and to be fair, after a couple of breathless paragraphs, the press release calms down a little and gets a bit more realistic. ‘Accuracy’ turns out to be ‘sensitivity’ of 90%; the specificity was 80%, the population was adult males from an autistic spectrum disorder clinic, and so on. I have since used these statistics as part of a short session that I run with doctors in training on understanding what a positive result really means, especially in unscreened populations. I have heavily ripped part of this off from a column by Ben Goldacre on the difference in meaning of a positive test in populations of high and low prevalence (which you can find here: http://bit.ly/CrystalBad), and preface the talk with the statement: ‘In 20 min you will be better at statistics than the press office of the MRC.’ It is always worth re-running these calculations yourself; I dare you to do it with a pencil and paper at the bottom of this page; work out what a positive scan result means in the context of an unscreened population with one in a hundred people having autism. Done it? Good: You should have got that a positive result means that you are a little less than 5% likely to have the diagnosis. We use this as a jumping off point to discuss how you should only do tests on carefully screened populations.
But back to the Pickets on bad papers. What to do with them? Shouldn't you be told when something is being badly misrepresented in the media? We thought so, but did not think we should label this as a Picket, because we want a Picket to represent something that has been done well, and which might alter or enhance our practice. So, the Autism Picket in this edition is a first and a last. Rather than invent yet another section for education and practice we are going to blog about them, via Bob Phillips' Archimedes blog, in the category ‘not-picket’. You can find them here: http://bit.ly/NotPicket.
The rest of this month has a very adolescent feel to it, and my editor's choice is a condition I find medically very challenging: The Medical Management of Acute Severe Anorexia Nervosa (see page 48).
Lastly, sorry for a few typos in the February edition; we are still getting used to the new format. My favourite was the one pointed out by Peter Ehrhardt who wondered, when we labelled the EEG for typical absence seizures as ‘3 Hz/s’, whether we should worry about acceleration of the seizures … Think about it – it is a nice geeky joke; I look forward to the groans when you get it.