george davey smith Profile picture
Apr 23 4 tweets 2 min read
1/n In 2019 @BMJ_latest published a non-linear Mendelian randomization (NLMR) paper with the conclusion that gaining weight would reduce mortality risk in many people. The NLMR method produces spurious findings MR's credibility calamity in this 🧵 bmj.com/content/364/bm…

Image
Image
2/n The sad story of NLMR and the delays in journals and authors in issuing retractions of clearly spurious papers containing damaging public health messages that include *literally impossible* findings is told here /cont
Mar 11 4 tweets 1 min read
Just seen that @ucl have made Lyndsay Farrall's 1969 PhD "The origins and growth on the English Eugenics movement" easily found with new intro by @profjoecain It remains perhaps the best overview of the topic, and deflates some myths. The survey of /contucl.ac.uk/sts/sites/sts/… 2/n the well-known members of the "Eugenics Education Society" and a random sample show academics and medical doctors the most common professions, that leftists as well as conservatives well represented, that Lamarckian eugenics was a thing (so much for the bizarre notion /cont
Jan 18 9 tweets 3 min read
A brilliant 1984 chapter in the "Encyclopaedia of Medical Ignorance" by Richard Peto anticipated much of the advances in understanding of cancer since then, and is now accessible in @SpringerNature European Journal of Epidemiology. A personal🧵on it/ cont link.springer.com/article/10.100… 1/n I first read this sitting on a stool in the HK Lewis medical bookshop on Gower Street soon after it came out; the book was outside my budget & the rest not so interesting. It felt like the 19th century in Lewis' anachronistic shop, there were hardly any customers /cont Image
Dec 11, 2023 10 tweets 3 min read
A twee story on tea by @TorstenBell in @guardian suggests increased consumption of tea (using boiled water) led to a fall in mortality in Britain. This might be true, but there is no mention of how this was linked to massively increased deaths in India 🧵
theguardian.com/commentisfree/…
Image 1/n The academic paper @guardain story is based on suggests that growing tea consumption by the working class after 1784 in Britain led to a fall in mortality. It uses what it calls "The Tea and Windows Act of 1784" which led to a dramatic fall /cont repec.iza.org/dp15016.pdf
Dec 1, 2023 12 tweets 5 min read
In 2018 we published a non-linear Mendelian randomisation (NLMR) study in @ObesitySociety and it is now clear the method we applied is deeply flawed, so in October we published a Perspective correcting it. A 🧵on the cautionary tale of NLMR onlinelibrary.wiley.com/doi/epdf/10.10…
Image 2/n Naïve observational studies report J- or U- shaped upturns of mortality at lower levels of BMI. Our paper using UK Biobank (UKB; @uk_biobank) data suggested overall adverse effects of BMI on mortality, and (though very imprecisely estimated) the NLMR suggested a lower /cont
Nov 3, 2023 8 tweets 3 min read
Ultra-processed foods (UPFs) are apparently the new poison, and a recent @bmj_latest report by @rebeccacoombes suggests conflicts of interests in the “expert” views presented @smc_london that weren’t enthusiastic enough in promoting this notion. /contbmj.com/content/383/bm… 2/n Opposing these @DoctorChrisVT says there is “overwhelming evidence” of UPFs being harmful. The prospective studies (contrary to the quote, these were not “trials”, where people were randomised to higher or lower UPF consumption) do not show UPFs cause harm / cont Image
Dec 7, 2022 7 tweets 3 min read
China is abandoning its zero covid before the policy further destabilizes the country. In the UK as well as China a self-appointed group denied the scientific evidence, which is that there was no other prospect than SARS-CoV-2 becoming endemic /cont
ft.com/content/b0c2f6… 2/n Once data on the transmission characteristics of SARS-CoV-2 became clear in Feb 2020 there was no other possible outcome than endemicity, explained clearly here by John Edmunds on Channel 4 News on 13th March 2020 /cont
Jul 22, 2022 12 tweets 6 min read
Great piece by @aoifemcl “The deceptive simplicity of mendelian genetics” in @PLOSBiology in which she says “I could not imagine trying to teach genetics without starting with Mendel” A short 🧵linking to discussion #Mendel200 by @KampourakisK et al (1/n) journals.plos.org/plosbiology/ar… 2/n Mendel’s laws - actually first formulated in the way now taught by Robert Heath Lock in 1906: biodiversitylibrary.org/item/68277#pag… are a useful heuristic, relating to processes that are likely essential for complex life to exist
Aug 11, 2021 12 tweets 5 min read
Can cognitive and other non-financial biases count as conflicts of interest (COIs)? COIs seem to be thought of as only absolutely direct financial conflicts – e.g. you receive tobacco industry money (££ to you or grants) whilst studying the health effects of smoking. A thread/ 2/n There are many cognitive biases – ranging from strong commitment to a particular theory through to membership of groups benefitting in ways that are not directly financial – that could influence what policy initiative or apparent scientific advance one puts forward / cont
Oct 1, 2020 5 tweets 2 min read
Covid cases are lowest in places that at the peak had the fewest or the most covid deaths. @dannydorling Tony Brookes and I considered reasons for this

theconversation.com/why-are-corona… Image 1/n The pattern could be a chance quirk, or could reflect testing regimens being related to previous severity of the epidemic, or safer behavioural patterns in areas badly affected, or more homeworking being possible in such places (or a mix of the above)
Sep 23, 2020 4 tweets 2 min read
1/n As usual several emails offering “acceptance within a week” for my submitted papers. I want the 24 hr European Journal of Clinical Investigation (Editor: John Ioannidis) service for a 27 page paper, senior author: John Ioannidis onlinelibrary.wiley.com/doi/epdf/10.11… Image 2/n The paper was a method for identifying "highly influential biomedical researchers". This approach suggested John Ioannidis was amongst the very most influential Image
Sep 22, 2020 18 tweets 5 min read
Very informative @jimalkhalili #TheLifeScientific with @neil_ferguson. A thread disputing his conclusion that we face "a trade off between saving lives and saving the economy and jobs". It is so much broader than that .. bbc.co.uk/sounds/play/m0… 1/n “lives vs economy” is unhelpful, it is not primarily the economy and jobs that are diminished by full or partial lockdowns, rather the almost unimaginable richness of life in all its domains suffers. We are social beings, and a major part of that sociality is stripped from us
Aug 12, 2020 5 tweets 2 min read
Great to see @klarawanelik @Zen_of_Science paper on how ethnicity, gender, sexual orientation and social background create barriers to academic career progression has now been published. Social origins under-represented in research and policies (.. cont/)
onlinelibrary.wiley.com/doi/epdf/10.10… 2/n perhaps partly due to social position obviously being influenced by educational trajectory. A possible measure would be whether parents left education at the minimum legal age of school leaving prevalent when and where they grew up. This would be calculable for many ...
Jun 23, 2020 13 tweets 5 min read
Below is the start of a thread by a geneticist on discovering that the most extreme single genetic variant he can find in @GWASCatalog actually tells you about the huge importance of the environment in causing disease ALDH2 variation strongly influences alcohol consumption, and can be used in Mendelian Randomization (MR) studies to demonstrate that alcohol drinking has a large unfavourable effect on blood pressure journals.plos.org/plosmedicine/a…
Jun 3, 2020 5 tweets 4 min read
Some surprise that @NEJM and @TheLancet both duped into publishing highly dubious hydroxychloroquine papers, but should we be surprised? .. (continued) theguardian.com/world/2020/jun… Probably not, both @TheLancet and @NEJM like high profile and apparently breakthrough studies. Both published fraudulent studies from the same Norwegian group on head and neck cancer thelancet.com/action/showPdf…
nejm.org/doi/pdf/10.105….
May 3, 2020 9 tweets 4 min read
Must read piece on cholera and COVID by @NeilSinghHQ A brief thread on whether it's true that in Broad Street John Snow “had the handle of the pump removed which led to a sharp decline in cholera cases in Soho”.
theguardian.com/society/2020/m… 1/n When Snow had the Broad Street Pump handle removed the outbreak was almost over.
academic.oup.com/ije/article/31… Image
Apr 28, 2020 7 tweets 2 min read
A brief thread on epidemiological issues when interpreting a recent study from Wuhan as perhaps indicating a 10% “population level antibody prevalence” .. it could be even higher 2/7 These data come from testing people applying to resume working after the lockdown. Details are scanty, but the key figure of 10% comes from the passage below onlinelibrary.wiley.com/doi/10.1002/jm… Image
Feb 26, 2020 6 tweets 4 min read
18th Feb issue of @JAMA_current has 3 interesting genetic papers which demonstrate the paradox that genetic data may tell us more about population than individual risk. Thread follows jamanetwork.com/journals/jama/… @JAMA_current Two papers demonstrate that - at the level of the individual patient - adding polygenic scores (PGS) for coronary heart disease (CHD) risk does not usefully improve prediction over conventional measures jamanetwork.com/journals/jama/… jamanetwork.com/journals/jama/…
Feb 22, 2020 4 tweets 4 min read
OK #epitwitter we have a winner for the quiz on what is the noun for "a residually confounded association of no causal significance" that Shah and I thought of 20 yrs ago but didn't print: it is @Lyonpaul Some reflections follow @Lyonpaul I thought @Lyonpaul falls into this category, but he denies it. I doubt he is denying he is never been in a pub with Shah and me and all slightly worse for wear? So he must think his memory is so fabulous that he can confidently say this was not said🤠
Jan 20, 2020 5 tweets 4 min read
So here's the thing .. @causalinf suggests "that Angrist and Krueger said that the best instruments came from deep institutional knowledge. Meaning, it comes from being intimately familiar with some phenomena" @yudapearl replies it's very simple to spot an IV in your model .. @causalinf @yudapearl and @yudapearl gives his paper as support ftp.cs.ucla.edu/pub/stat_ser/r… saying that such are "homework problems in Economics 101 class". This is based on identifying an IV **in your DAG** ..
Jan 5, 2020 25 tweets 14 min read
Gr8 @EpiEllie “Sick individuals and sick populations” (SISP) a classic; along with his many other writings (including his valedictory book “Strategy of Preventive Medicine”) it should have remade the epidemiological landscape. A thread here on a few interpretive differences @EpiEllie 1/n His views don’t fit within the “well defined interventions” envelope. His central idea is that the causes of health differences between populations and between individuals can be different. Consider his oft-used figure on circulating cholesterol levels in Japan and Finland