Language, Truth and Logic
“Perhaps tellingly, it wasn’t easy finding someone to argue for the benefits to health in leaving the EU, but Graham Gudgin agreed.” Crisp, punchy copy; but misleading and factually incorrect.
Richard Hurley, the BMJ’s features and debates editor, offered me the 600 word debate on the 24th July. We agreed that my response would wait until I returned from holiday. On the 14th August, for reasons outlined below, I declined, instead suggesting Dr Graham Gudgin. He was approached on the 17th August and accepted on the 19th.
Tellingly or not, ‘finding’ doesn’t come much easier than that.
The offer of a debate followed my criticism of an editorial (1) which put only one side of a case. I bemoaned, not for the first time (2), the lack of external refereeing in editorials and commentaries where the BMJ gave advice on matters of health. My complaint was that the Journal was breaking many of its own rules on the handling of evidence (evidence-based medicine; open data; partnering with patients) and included a predominance of confirmation-biased references (3).
Consistently, and purposely, I have made no case for or against Brexit. For that reason, suggesting I should debate in favour of Brexit was inappropriate. In the lengthy published to-and-fro (and additionally with Richard Hurley and Fiona Godlee), I have repeatedly stressed that I was urging the use of an independent referee (I spare the dates and quotes for any of you who are still awake); I was NOT arguing for or against leaving the EU (I happen to be for). I suggested that, since the original editorial had relied on questionable and unreliable financial data and conclusions, an economist would be better suited; Dr Gudgin was easily found.
As far as statistics go, we are all familiar with their relationship to lies and damned lies:
Of the 46,500,001 UK electorate in 2016, 29,089,259 (62.6%) did not vote to leave (4).
Equally truthfully, 30,358760 (65.3%) did not vote to remain. Both quotes are needed for balance.
“Around 2500 EU nationals (net) joined the NHS in the quarter before the referendum. Between March and December 2017 the figure was just over 200”. (5)
Rebecca Coombes was partial when she chose to repeat Anand Menon’s numbers which “plummeted” from 2,500 to 200, while ignoring Graham Gudgin’s 7,000 (actually 6,888) increase in new EU recruits in the two years to January 2018 (6). Both Menon’s and Coombes’ links (7, 8) are opaque (even NHS Digital (9) were unable to replicate Coombes’ figures), but Gudgin’s (10) are easy to verify (pace A J Ayer). What is clear is that Menon, apart from using different denominators for his figures (3 and 10 months), is selective in his choice.
I did, eventually, find a source with quarterly figures (11). Access them and see that Menon prefers his 200 (actually, 365 (62,299 – 61,934), and actually not March, but April to December), to the 1,916 (net) increase for the calendar year 2017 and the rather less doom-laden 957 increase in the first three months of 2018. Apart from debatable arithmetic and English, cherry-picked numbers have been used to advance his argument.
Suppose a statin’s formulation had been changed in June 2016, and identical figures to the above were offered demonstrating the diminution in side-effects before and after. The BMJ would have rightly examined the figures minutely and lambasted the drug company for crooked, dishonest dealing.
There is no consistency in the BMJ’s approach to data and evidence; and no logic in having a stringent set of guidelines for the original research section which are then ignored in the early pages. I suggest again that, when editorials involve medical decisions, external referees would improve both content and credibility. Obviously the large majority of news and views need only be scanned in-house but I do expect a balanced and scientific argument in both research pieces and editorials.
In this instance, why use ‘tellingly’ (OED = significantly) when a brief word with an office colleague would have revealed the facts? Why, when Menon’s link to NHS Digital was so obscure, did Coombes repeat this opaque link while misquoting him – his denominator for 200 was 10 months not three? The importance of authors checking their references and sources has been emphasised by the BMJ (12). Did the author access the tables and, if so, why not use a clearer link (10) or even check the numbers herself? Did anybody in the Office read the references?
I feel uncomfortable prolonging this squabble and happily emphasise that correspondence has at all times been polite and gracious but at no time has anyone addressed my original complaint that, without independent refereeing, the BMJ puts at risk its hard earned reputation in dealing with evidence.
1. Gill M, McKee M, Malloch Brown M, Godlee F. Brexit is bad for our health. BMJ2018;361:k2235 . doi:10.1136/bmj.k2235 pmid:29794034
2. Barlow D. Bias in medical literature on health outcomes; bias in commentary?
BMJ 2016; 355 doi: http://dx.doi.org/10.1136/bmj.i6634 (Published 16 December 2016)
3. Barlow D. Brexit - an impartial view? BMJ 2018; 361 doi: https://doi.org/10.1136/bmj.k2743
4. Electoral Commission https://www.electoralcommission.org.uk/find-information-by-subject/elect...
5. Menon A. Could Brexit harm the NHS? Yes! https://doi.org/10.1136/bmj.k4014
6. Gudgin G. Could Brexit harm the NHS? No! https://doi.org/10.1136/bmj.k4014
7. NHS Digital. NHS workforce statistics, December 2017. https://digital.nhs.uk/data-and-information/publications/statistical/nhs...
9. NHS Digital Contact Team/Workforce Analysis Team. Personal communication Oct 29. ref NIC-241581-D9Q6P
11. https://files.digital.nhs.uk/3C/F5CB38/HCHS%20staff%20in%20NHS%20Trusts%... (published 21 June)
12. Godlee F, Smith Jane, Marcovich H. Wakefield’s article linking MMR vaccine and autism was fraudulent; BMJ 2011;342:c7452
Competing interests: My long-standing, happy, and continuing connections with Europe are detailed elsewhere (3). I voted to leave the EU. In 2015 my name was added to the BMJ’s panel of doctors prepared to work on conflict of interest.