3 ms·

The ONS data cannot be used that way unfortunately, and the ONS themselves have admitted to that fact. If you take the ONS data literally then COVID vaccines ar

The ONS data cannot be used that way unfortunately, and the ONS themselves have admitted to that fact. If you take the ONS data literally then COVID vaccines are a magical elixir of luck that protect against every cause of death including car crashes. There are at least two problems: 1. The statistically invalid time-windowing games the public health agencies all played in which people who had taken vaccines were classed as unvaccinated. The sibling comment talks about that. Prof Norman Fenton has shown how this yields incorrectly high calculations of effectiveness. 2. Healthy vaccinee bias, in which people who are about to die aren't vaccinated at all because there's no point, and the sort of people who take lots of vaccines tend to be obsessive about health and risk in other ways. Problems like these with observational data are why drugs are put through trials before being launched to market. If we check the data for say the Pfizer trial, what we see is no effect on mortality and serious problems with statistical power also, simply because so few people were dying of COVID it was basically impossible to run a trial large enough to prove it had any effect.

kadkadels 3y ago
We are not talking about vaccine efficiency or efficacy here though but adverse events specifically. >If we check the data for say the Pfizer trial, what we see is no effect on mortality The point is that there are almost no serious adverse events occuring post-vaccination. There is no pharmacovigilance safety signal or concerning pattern wrt MRNA-vaccines. As we have seen in 2021 with the astra zeneca vaccine, post market surveillance worked and it was promptly investigated as soon as a safety signal emerged. EMA (and other regulatory agencies) communicated transparently about their findings and decisions. Thank you for reminding me of this fine example of post-market surveillance, I almost forgot :)
pw201 3y ago
> The statistically invalid time-windowing games the public health agencies all played in which people who had taken vaccines were classed as unvaccinated As I have just replied to the other commenter, the ONS data he appears to object to categorises various "vaccinated" categories starting immediately after vaccination. The regulator's reply to Fenton makes this clear: https://osr.statisticsauthority.gov.uk/correspondence/ed-humpherson-to-norman-fenton-martin-neil-clare-craig-and-scott-mclachlan-ons-deaths-by-vaccination-status-statistics/ https://osr.statisticsauthority.gov.uk/correspondence/ed-hum... I assume this reply is what you refer to when you say that the ONS admit the data cannot be used that way. However, since that reply, the ASMR calculation now uses data linked to the 2021 census which covers around 91% of the population. Paul Mainwood graphed the ASMRs here: https://twitter.com/PaulMainwood/status/1627979309812965381 https://twitter.com/PaulMainwood/status/1627979309812965381 I see no evidence here that being vaccinated makes you more likely to die, which is what the original thread was about.
- mike_hearn 3y ago
  Humpherson simply says that "ONS ... do receive data in [people within two weeks of vaccination], and that the individual would fall into the ‘vaccinated’ category" which is a meaningless tautology, as the entire argument is about the definition of vaccinated used by the public health agencies. Invalid definitions are a very real and pervasive problem, not only being standard in the public health literature but also standard practice for public health agencies. Fenton has compiled a partial (!) list of papers and studies which do this: https://wherearethenumbers.substack.com/p/the-very-best-of-cheap-trick https://wherearethenumbers.substack.com/p/the-very-best-of-c... Note that UK HSA and studies using ONS data are shown to use invalid definitions. > the ASMR calculation now uses data linked to the 2021 census which covers around 91% of the population They say it does, but the UK government doesn't really the true population size. They've admitted this in the past. The 2021 census is unlikely to have solved it given that the civil service doesn't really want to know (if they could actually get full coverage, then they could identify every illegal immigrant). This showed up badly during COVID where more people came forward to be vaccinated than theoretically existed at all in certain age groups. They also told people that only a small percentage of the population was unvaccinated, but when the BBC commissioned a poll as part of a programme they were making (on misinformation!) around 25% of respondents said they were unvaccinated, a huge gap. So the data quality here is of garbage grade in almost every respect, which is a scandal. People who just tune out and decide nothing governments say on the topic can be trusted are well within reasonable bounds.
  - pw201 3y ago
    > which is a meaningless tautology, as the entire argument is about the definition of vaccinated used by the public health agencies. The ONS mortality stats linked to by kadkadels at the start of the thread contain data related to people who received at least one COVID vaccine (counted from the day they received it, subdivided by time since reception and number of boosters), as well as an unvaccinated category who never did. This is clear to anyone who read the link that kadkadels posted. Both you and armchairdweller falsely claimed that the unvaccinated category included people who received the vaccine less than N days ago, presumably because you believe that some deaths caused by the vaccine shortly after people receive it are hidden by these stats. But in fact, the vaccinated category starts from the day of vaccination, the unvaccinated tended to die more back when COVID was new, and the ASMR for unvaccinated and vaccinated converged by about the end of 2022, presumably because we've nearly all had COVID at least once so being vaxed now isn't doing a whole lot of good. Both the "vax did more harm than good" crowd and the "we should still be wearing masks" crowd are wrong. Fenton is in HART, and HART are off their collective rockers, as we knew pretty early on when their internal chats were leaked. See https://www.logically.ai/articles/hart-files-anti-vaccine-myths-westminster https://www.logically.ai/articles/hart-files-anti-vaccine-my... and https://twitter.com/_johnbye/status/1421397013078360064 https://twitter.com/_johnbye/status/1421397013078360064 for example, and my own small part in pointing out that all the astroturfing groups identified by Neil O'Brien MP were hosted on a single IP address (HART almost immediately moved, lol): https://twitter.com/nameandnature/status/1352998804832870402 https://twitter.com/nameandnature/status/1352998804832870402 Though the prime mover, Narice Bernard, seems to have moved on to newer conspiracies involving "climate lockdowns" and "15 minute cities", people who conclude that nothing HART say on COVID topics can trusted are well within reasonable bounds.

kadkadels 3y ago

We are not talking about vaccine efficiency or efficacy here though but adverse events specifically. >If we check the data for say the Pfizer trial, what we see is no effect on mortality The point is that there are almost no serious adverse events occuring post-vaccination. There is no pharmacovigilance safety signal or concerning pattern wrt MRNA-vaccines. As we have seen in 2021 with the astra zeneca vaccine, post market surveillance worked and it was promptly investigated as soon as a safety signal emerged. EMA (and other regulatory agencies) communicated transparently about their findings and decisions. Thank you for reminding me of this fine example of post-market surveillance, I almost forgot :)

pw201 3y ago

> The statistically invalid time-windowing games the public health agencies all played in which people who had taken vaccines were classed as unvaccinated As I have just replied to the other commenter, the ONS data he appears to object to categorises various "vaccinated" categories starting immediately after vaccination. The regulator's reply to Fenton makes this clear: https://osr.statisticsauthority.gov.uk/correspondence/ed-humpherson-to-norman-fenton-martin-neil-clare-craig-and-scott-mclachlan-ons-deaths-by-vaccination-status-statistics/ https://osr.statisticsauthority.gov.uk/correspondence/ed-hum... I assume this reply is what you refer to when you say that the ONS admit the data cannot be used that way. However, since that reply, the ASMR calculation now uses data linked to the 2021 census which covers around 91% of the population. Paul Mainwood graphed the ASMRs here: https://twitter.com/PaulMainwood/status/1627979309812965381 https://twitter.com/PaulMainwood/status/1627979309812965381 I see no evidence here that being vaccinated makes you more likely to die, which is what the original thread was about.

mike_hearn 3y ago
Humpherson simply says that "ONS ... do receive data in [people within two weeks of vaccination], and that the individual would fall into the ‘vaccinated’ category" which is a meaningless tautology, as the entire argument is about the definition of vaccinated used by the public health agencies. Invalid definitions are a very real and pervasive problem, not only being standard in the public health literature but also standard practice for public health agencies. Fenton has compiled a partial (!) list of papers and studies which do this: https://wherearethenumbers.substack.com/p/the-very-best-of-cheap-trick https://wherearethenumbers.substack.com/p/the-very-best-of-c... Note that UK HSA and studies using ONS data are shown to use invalid definitions. > the ASMR calculation now uses data linked to the 2021 census which covers around 91% of the population They say it does, but the UK government doesn't really the true population size. They've admitted this in the past. The 2021 census is unlikely to have solved it given that the civil service doesn't really want to know (if they could actually get full coverage, then they could identify every illegal immigrant). This showed up badly during COVID where more people came forward to be vaccinated than theoretically existed at all in certain age groups. They also told people that only a small percentage of the population was unvaccinated, but when the BBC commissioned a poll as part of a programme they were making (on misinformation!) around 25% of respondents said they were unvaccinated, a huge gap. So the data quality here is of garbage grade in almost every respect, which is a scandal. People who just tune out and decide nothing governments say on the topic can be trusted are well within reasonable bounds.
- pw201 3y ago
  > which is a meaningless tautology, as the entire argument is about the definition of vaccinated used by the public health agencies. The ONS mortality stats linked to by kadkadels at the start of the thread contain data related to people who received at least one COVID vaccine (counted from the day they received it, subdivided by time since reception and number of boosters), as well as an unvaccinated category who never did. This is clear to anyone who read the link that kadkadels posted. Both you and armchairdweller falsely claimed that the unvaccinated category included people who received the vaccine less than N days ago, presumably because you believe that some deaths caused by the vaccine shortly after people receive it are hidden by these stats. But in fact, the vaccinated category starts from the day of vaccination, the unvaccinated tended to die more back when COVID was new, and the ASMR for unvaccinated and vaccinated converged by about the end of 2022, presumably because we've nearly all had COVID at least once so being vaxed now isn't doing a whole lot of good. Both the "vax did more harm than good" crowd and the "we should still be wearing masks" crowd are wrong. Fenton is in HART, and HART are off their collective rockers, as we knew pretty early on when their internal chats were leaked. See https://www.logically.ai/articles/hart-files-anti-vaccine-myths-westminster https://www.logically.ai/articles/hart-files-anti-vaccine-my... and https://twitter.com/_johnbye/status/1421397013078360064 https://twitter.com/_johnbye/status/1421397013078360064 for example, and my own small part in pointing out that all the astroturfing groups identified by Neil O'Brien MP were hosted on a single IP address (HART almost immediately moved, lol): https://twitter.com/nameandnature/status/1352998804832870402 https://twitter.com/nameandnature/status/1352998804832870402 Though the prime mover, Narice Bernard, seems to have moved on to newer conspiracies involving "climate lockdowns" and "15 minute cities", people who conclude that nothing HART say on COVID topics can trusted are well within reasonable bounds.