Friday, 5 June 2020

Might be cobblers

Only a week ago I was all excitey about a massive study which showed that hydroxychloroquine HCQ, much pushed by POTUS, was more likely to kill people than cure them. As you do, I accepted that the data were okay and had a cursory look at the statistical analysis and more-or-less accepted the conclusions of the paper. That was a Convenient Truth because, after 3½ years of political iconoclasm, I think I prefer old-style, slightly boring, statesmanlike statesmen rather than comedians, hacks and braggarts. I could / should have polished my crap-detector rather than the bottled genie because I've had to flag the old post with this new disclaimer:

</bollix Alert> 04Jun20: The paper which triggered this post has been flagged EOC by the Lancet. EOC Expression of Concern is a not-quite[-yet-]retraction. The Grauniad has investigated the company which supplied the data on which the original study was based. Also check out commentary on Metafilter</bollix alert>

The Grauniad, which would probably be with me on who should be POTUS, was less invested in the HCQ=Bad than I was. They noted out that on 21st April, the "Lancet dataset" was reporting a cumulative Australian death-count of 73 whereas JHU was claiming only 67. Now that could have been a date mix-up because there were 6 extras deaths in the following two days OR an Asian hospital could have been inadvertently included in the Australian dataset. This last plea comes from the Database Company's Response.  But, smelling something fishy, The Grauniad's terriers started to unravel some other threads in the dataset and/or the credentials of the people who had assembled it.  It's a bit like applying the Fodor's guide test to it: if you easily find an error or inconsistency you push a bit harder on other elements of the edifice.

In the lynch-mob spirit of duffing up The Lancet, its editor, the paper's referees, I'll remind you that The Lancet published Andrew Wakefield's 1998 paper which set out the antivax stall that MMR ⇨ ⇨ autism. They are cranking up their reaction time at The Lancet because the current EoC flicked up 12 days after they published the original paper. It took them 12 years to slap a 2010 retraction on Wakefield's study. It's a problem that far fewer people read retractions than the original splashy paper; especially if they are invested in the results. On Wakefield, which had a sample size of 12, you could have wished for the Lancet to be more "extraordinary claims require extraordinary levels of proof". The current paper ironically had numbers too big to be credible. Call Dr Goldilocks, she knows all the best bars reasonable datasets.

That "extraordinary claims  . . ." is an autoquote from an earlier Blob which riffed on how easy it is to publish rublish. Even Science, the premier general science magazine West of the Mid-Atlantic Ridge, had egg on its face over an incredible paper suggesting a form of life that managed without phosphorous a key ingredient of DNA. One of the targets of Trumpian populist spite is the WHO. The World Health Organisation is an intergovernmental international quango with jobs-for-the-boys for more than 7000 people from more than 150 countries. In these pandemic times it is vital to have a single authority and coordinator of medical, epidemiological responses to health crises. It is large but not beyond criticism and could certainly have some of its fat trimmed off. But we are far better off with the WHO and all its warts than without it and fighting ineffective, fragmented,Them/US battles against HIV SARS MERS Ebola Zika CoVid-19 Cadbury-99. Not unlike the EU; [necessary] criticism of which has decidedly toned down through these Brexit years.

In response to a single paper in The Lancet, the WHO abruptly stop shuttered its own investigation of the efficacy and safety of HCQ in treating CoVid-19. And then abruptly go started it up again in response to a single article (and the sound of  fffrrrrrPPP hitting the fan from other directions) from the Guardian. Agile reaction times are an asset in any large organisation but we could hope for something a little more magisterial and considered in its decisions. If, despite the buffets and nit-pickets of the pack of critics, the Lancet paper is found to be substantively correct, will WHO stop their study again?  On again off again begin again Finnegan.

No comments:

Post a Comment