3 ms·

The linked video, https://www.youtube.com/watch?v=Hc7A1bVuSJU&t=56s https://www.youtube.com/watch?v=Hc7A1bVuSJU&t=56s, is misleading. It shows that the number o

The linked video, https://www.youtube.com/watch?v=Hc7A1bVuSJU&t=56s https://www.youtube.com/watch?v=Hc7A1bVuSJU&t=56s, is misleading. It shows that the number of unvax cases is much larger than vax cases. I hope that by know it is common knowledge that the pandemic behaves very differently between age groups. Simpson's paradox abounds. The video only breaks down vax/unvax for aggregate numbers. This is a smell. UK publishes the vax/unvax numbers broken down by age groups. See Table 2 in https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/1031157/Vaccine-surveillance-report-week-44.pdf https://assets.publishing.service.gov.uk/government/uploads/.... Raw numbers: cases between week 40 and week 43 2021 unvax vax 1dose under 18 336893 818 36000 18-29 24097 33649 9500 30-39 25832 71492 7800 40-49 15717 130686 4500 50-59 6701 99722 2000 60-69 2484 56779 800 70-79 917 34305 300 ≥80 471 12339 200 1. In every age group, except under 18, the number of vax cases is larger than the number of unvax cases, contradicting the video on its face. 2. Under 18 has a large case number, which dominates the totals. Some issues: 2.1. Schools have strict testing and tracing policies. The large case numbers may simply an effect of higher testing rates for the respective population. 2.2. Related to 2.1., there are two ways to count cases: PCR cases and symptomatic cases. In case of large case numbers outliers, the difference between the two measures can explain a large difference, though it is impossible to independently verify that a consistent measure was applied for all age groups. 2.3. Vaccination in 12-17 was only recently introduced. VE decreases over time. Studies disagrees by how much. At six months time, for Pfizer I've seen 55%-75%, from a high of 90%. Closer to the vaccination event VE is higher, thus the number of vax cases is temporarily lower. A more realistic assessment assessment ought to a. break down under 18 in to 0-5, 5-11, 12-17, and wait 6 months post vaccination campaign. Note that VE is also a somewhat misleading metric, 90% means 9/10 case reduction, 50% merely 1/2; a small percentage drop in VE at the high values mean a large increase in cases. 3. In some age categories, e.g. 50-59, unvax share of the cases is as small as 6%. Not sure (*) what the unvax rate in UK is for that age group, the reports says somewhere in the low teens, that number has been contested. But 6% is low enough by itself. If that doesn't make you wonder about the official line "unvaccinated people were 6.8 times more likely to get COVID-19 than fully vaccinated people.", then nothing will ever do. Can't speak about US data because US is, for some reason, very tight lipped when it comes to publishing basic epi data with breakdown by date, age and vax status. If you have a link to the equivalent of the UK PHE vaccine surveillance report for any US state, please share. (*) Apparently UK is not sure about the vax rates either, because apparently UK doesn't know how many people of a given age live in the country. That is a story for another time.

SideburnsOfDoom 5y ago
Thanks for linking to that report, it's interesting reading, particularly the part with "We present data on COVID-19 cases, hospitalisations and deaths by vaccination status. These raw data should not be used to estimate vaccine effectiveness" (that sentence occurs twice so tat you don't miss it) This is of course exactly what you did above. It also says "We have published a blog to accompany this section of the vaccine surveillance report" https://ukhsa.blog.gov.uk/2021/11/02/transparency-and-data-ukhsas-vaccines-report/ https://ukhsa.blog.gov.uk/2021/11/02/transparency-and-data-u... You should not expect ZOE study data to match UK Govt data exactly, they use different methods and check different symptoms. This is that theme that Prof. Spector talks about often.
- angelzen 5y ago
  "should not be used to estimate vaccine effectiveness" Exactly. Neither should aggregate numbers that various US states and Prof. Spector use to spread fear, acrimony and stigma should be used to estimate VE, or mislead the audiences to estimate VE by themselves. Notice how the ZOE video you linked lacks the intellectual honesty to display the same disclaimer. His own website says "Incidence estimates were updated on 21 July 2021.". He's basically fabricating numbers, for all we know just as accurate as Prof. Ferguson. This is no better than the "antivax garbage" some people are so keen on summarily calling out. Here are some longish term studies and preprints for Pfizer's VE against infection. UK, 3 months VEaI = 75%. https://www.nature.com/articles/s41591-021-01548-7 https://www.nature.com/articles/s41591-021-01548-7 Qatar, 3 months VEaI = 70%, 6 months VEaI = 12.8%, https://www.nejm.org/doi/full/10.1056/NEJMoa2114114 https://www.nejm.org/doi/full/10.1056/NEJMoa2114114 Sweden, <50yo, 2-4 months VEaI = 83%, 4-6 months VEaI = 50%, https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3949410 https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3949410 US, veterans, 3 months VEaI = 78%, 6 months VEaI = 50%, https://www.medrxiv.org/content/10.1101/2021.04.20.21254636v1 https://www.medrxiv.org/content/10.1101/2021.04.20.21254636v... Two notes: * VE against infection drops over time. Sadly, I am not aware of any reliable literature for VE after 9 or 12 months. Unfortunately Pfizer killed their own critical trials, which had a solid headstart, by vaccinating the control arm after 6 months. But the trend is clear, and worrisome. * 80% VE => 1:5 infection odds. 75% VE => 1:4 infection odds. 50% VE => 1:2 infection odds. 12% VE, it's a wash. Numbers like 1:6.8 infection odds this long after the vaccination campaign are fantasy.

SideburnsOfDoom 5y ago

Thanks for linking to that report, it's interesting reading, particularly the part with "We present data on COVID-19 cases, hospitalisations and deaths by vaccination status. These raw data should not be used to estimate vaccine effectiveness" (that sentence occurs twice so tat you don't miss it) This is of course exactly what you did above. It also says "We have published a blog to accompany this section of the vaccine surveillance report" https://ukhsa.blog.gov.uk/2021/11/02/transparency-and-data-ukhsas-vaccines-report/ https://ukhsa.blog.gov.uk/2021/11/02/transparency-and-data-u... You should not expect ZOE study data to match UK Govt data exactly, they use different methods and check different symptoms. This is that theme that Prof. Spector talks about often.

angelzen 5y ago
"should not be used to estimate vaccine effectiveness" Exactly. Neither should aggregate numbers that various US states and Prof. Spector use to spread fear, acrimony and stigma should be used to estimate VE, or mislead the audiences to estimate VE by themselves. Notice how the ZOE video you linked lacks the intellectual honesty to display the same disclaimer. His own website says "Incidence estimates were updated on 21 July 2021.". He's basically fabricating numbers, for all we know just as accurate as Prof. Ferguson. This is no better than the "antivax garbage" some people are so keen on summarily calling out. Here are some longish term studies and preprints for Pfizer's VE against infection. UK, 3 months VEaI = 75%. https://www.nature.com/articles/s41591-021-01548-7 https://www.nature.com/articles/s41591-021-01548-7 Qatar, 3 months VEaI = 70%, 6 months VEaI = 12.8%, https://www.nejm.org/doi/full/10.1056/NEJMoa2114114 https://www.nejm.org/doi/full/10.1056/NEJMoa2114114 Sweden, <50yo, 2-4 months VEaI = 83%, 4-6 months VEaI = 50%, https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3949410 https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3949410 US, veterans, 3 months VEaI = 78%, 6 months VEaI = 50%, https://www.medrxiv.org/content/10.1101/2021.04.20.21254636v1 https://www.medrxiv.org/content/10.1101/2021.04.20.21254636v... Two notes: * VE against infection drops over time. Sadly, I am not aware of any reliable literature for VE after 9 or 12 months. Unfortunately Pfizer killed their own critical trials, which had a solid headstart, by vaccinating the control arm after 6 months. But the trend is clear, and worrisome. * 80% VE => 1:5 infection odds. 75% VE => 1:4 infection odds. 50% VE => 1:2 infection odds. 12% VE, it's a wash. Numbers like 1:6.8 infection odds this long after the vaccination campaign are fantasy.