I've never heard the phrase "neighborhood deprivation" used. I know of depreciation, but I guess using such a word would make the evidence too obvious (those in lower socioeconomic areas are more likely to have poorer outcomes in general). I also don't provide these people with any charity in their abilities to completely butcher studies for whatever narrative they want to push forward.
And thanks for covering the statistics. I absolutely hate statistics and I just gloss over so much of that in papers.
Deprivation indexes seem pretty common in pregnancy studies, which makes sense, since the baby doesn't come with biomarkers for the mother's lack of care and so poverty is the most useful proxy for confounding risk. On the other hand, it probably obscures a lot of nuance. A teenaged high-DPI mother in Alameda county won't be the same as a 30-35 high-DPI mother in same. I say all this as a born "solid 4."
"Depreciation" is for accounting, hehe - but presumably babies also lose 10% of their value as soon as you drive them off the lot.
I consider The Cat to be mostly a muckraker (although I'm not positive that's the word I'm looking for). I've caught a few things where he misrepresents (or leaves out key information) on a subject or an event in order to agitate. But it does get lots of comments and it does keep readers agitated so - success! 😜
It's quite unfortunate because I hoped that Substack would actually encourage people to read more into studies. Unfortunately, I'm a bit disheartened in finding that many Substacks are turning into Twitter echo chambers rather than areas to inform people.
Thanks for pointing out the flaw in Gato's post. How did he possibly make that mistake???
It makes me worry about him a little.
What matters in this kind of analysis is the comparison between the pre booster rate (double vax) and the post-booster-pre-effect rate ("early prebooster"), which in this case are basically the same rate. So there's really no effect here. Who knows what's going on in the other days they excluded, that is a little fishy, but it's unlikely to be so bad as to make the shots ineffective... perhaps less effective though.
A fair way to calculate this would probably add the raw count of excess pre-boster events, that are over and above the pre-existing doublevax rate, to the raw booster counts, and use the original denominator for boosted days. That we we only put the extra events on boosting's tab.
A few people in his comments section also called this out.
I hope he retracts it, he retracted another post recently, although I didn't actually see a problem with it (I didn't look too hard though).
Work too long with corrupted data, break your disk. One reason why I never obsessed over the worry window and stopped futzing with UKHSA reports a long time ago.
“Sometimes” statistics can be deceiving; sometimes study authors make design decisions to account for this. I think that’s fair since I don’t care about statistics anyway and just want to see the raw data and make up my own mind; he seems to think that trying to correct for statistical deception is deception itself. That’s all that this worry window issue is really about. So for example in this one he seems to miss that rates of infections for the days after injection for every age group are clearly plotted in the supplemental (*edit: not even in the supplemental, right there in Fig 2), just as they were in the August Bar-On, et al. booster paper. I don’t know what the retracted “Israel” post was either, I wonder if it was the same paper. I noticed that lots of his commenters caught the error but their language was too math-y so I wanted to offer my simplified deconstruction.
I don't know much about statistics, so I depend on the substackers to provide some clarity. When I read the cat's article that you reference, I thought he was taking liberty with something that didn't have actual data behind it, so I took his article with a grain of salt.
I glossed through it without checking the work - I'm glad Brian checked it. I need to work on better confirmation bias avoidance. Big thank you to Brian!
BTW I agree with your idea not to obsess over the worry window. I think it actually does exist, but it's not incredibly significant as pertains to vaccine effectiveness in terms of individual outcomes. That is to say, there is a measurably higher risk in that window (this is what you find from the UK stats) but when amortized over the 6 month duration of effectiveness it's not really a big deal. (And the Gato post data didn't even find that result, although that could be because they only published part of the data for the worry window). At some point I did the math on the UK data and found IIRC about 2x rate increase over that short time window. For a person getting the shot it's pretty insignificant.
HOWEVER, it's not necessarily insignificant from the standpoint of an epidemic. If that effect causes a brief but substantial increase in infection rate among a significant group of people, it's basically increasing R0 in the short term, leading to spectacular effects in the case rates exploding during rollouts. Because that small rate increase has an exponential effect on case rate. And that in turn can feed back to overall greater harm. And people have also noticed this effect (leading to "the vaccine is the disease" which sometimes goes full tin foil).
The problem is that teasing these two concepts apart is harder to explain and model, so you get some shoddy analysis, leading to vax skeptics being "right" for the wrong reasons (two wrong arguments conflating effects that really are harmful).
I'm a subscriber so I have the israel post in email form if you are interested I can email it to you.
Your explanation was less mathy but maybe still more complicated than needed - if I understand it, his error was adding two fractions with different denominators. Even these days, this is apparently taught in 6th or 7th grade math class.
Yes, in mine bullets 1 and 2 are self-redundant. But they're both my internal "proofs" that I use so I don't have to actually inspect the math when someone says a change to an IRL denominator can change a per-x rate, haha
Thank You for this important post! Didn't Pfizer mention in their own first trial studies of their genetically modifying cocncoction, for the pregnant women to not even get close to the injected?? It is called shedding. I knew personally only one pregnant woman, and she lost her baby after being exposed to all injected grandparents! Here from a TOP lawyer about catastrophic consequences of the injections among the military personnel:
But the perfunctory nature of the protocol document should be kept in mind. It’s very important from a CYA standpoint that the text of the document is legally impeccable. It’s totally irrelevant as we have seen whether the text is actually practicable at scale or actually carried out by the trial vendors.
I've never heard the phrase "neighborhood deprivation" used. I know of depreciation, but I guess using such a word would make the evidence too obvious (those in lower socioeconomic areas are more likely to have poorer outcomes in general). I also don't provide these people with any charity in their abilities to completely butcher studies for whatever narrative they want to push forward.
And thanks for covering the statistics. I absolutely hate statistics and I just gloss over so much of that in papers.
Deprivation indexes seem pretty common in pregnancy studies, which makes sense, since the baby doesn't come with biomarkers for the mother's lack of care and so poverty is the most useful proxy for confounding risk. On the other hand, it probably obscures a lot of nuance. A teenaged high-DPI mother in Alameda county won't be the same as a 30-35 high-DPI mother in same. I say all this as a born "solid 4."
"Depreciation" is for accounting, hehe - but presumably babies also lose 10% of their value as soon as you drive them off the lot.
Thank you! I'm so glad when you all check each other's work and point out the weaknesses. We are all better for it. 👍🏽💕
Exactly, or rather, so you would think! Analyzing data without facing criticism and push-back is like forging a sword at room temp.
I consider The Cat to be mostly a muckraker (although I'm not positive that's the word I'm looking for). I've caught a few things where he misrepresents (or leaves out key information) on a subject or an event in order to agitate. But it does get lots of comments and it does keep readers agitated so - success! 😜
It's quite unfortunate because I hoped that Substack would actually encourage people to read more into studies. Unfortunately, I'm a bit disheartened in finding that many Substacks are turning into Twitter echo chambers rather than areas to inform people.
Which is funny since it’s not like there’s a lack of valid things to agitate people with! But people gravitate to the “negative efficacy: PROOF!” meme
Thanks for pointing out the flaw in Gato's post. How did he possibly make that mistake???
It makes me worry about him a little.
What matters in this kind of analysis is the comparison between the pre booster rate (double vax) and the post-booster-pre-effect rate ("early prebooster"), which in this case are basically the same rate. So there's really no effect here. Who knows what's going on in the other days they excluded, that is a little fishy, but it's unlikely to be so bad as to make the shots ineffective... perhaps less effective though.
A fair way to calculate this would probably add the raw count of excess pre-boster events, that are over and above the pre-existing doublevax rate, to the raw booster counts, and use the original denominator for boosted days. That we we only put the extra events on boosting's tab.
A few people in his comments section also called this out.
I hope he retracts it, he retracted another post recently, although I didn't actually see a problem with it (I didn't look too hard though).
Work too long with corrupted data, break your disk. One reason why I never obsessed over the worry window and stopped futzing with UKHSA reports a long time ago.
“Sometimes” statistics can be deceiving; sometimes study authors make design decisions to account for this. I think that’s fair since I don’t care about statistics anyway and just want to see the raw data and make up my own mind; he seems to think that trying to correct for statistical deception is deception itself. That’s all that this worry window issue is really about. So for example in this one he seems to miss that rates of infections for the days after injection for every age group are clearly plotted in the supplemental (*edit: not even in the supplemental, right there in Fig 2), just as they were in the August Bar-On, et al. booster paper. I don’t know what the retracted “Israel” post was either, I wonder if it was the same paper. I noticed that lots of his commenters caught the error but their language was too math-y so I wanted to offer my simplified deconstruction.
I don't know much about statistics, so I depend on the substackers to provide some clarity. When I read the cat's article that you reference, I thought he was taking liberty with something that didn't have actual data behind it, so I took his article with a grain of salt.
I glossed through it without checking the work - I'm glad Brian checked it. I need to work on better confirmation bias avoidance. Big thank you to Brian!
BTW I agree with your idea not to obsess over the worry window. I think it actually does exist, but it's not incredibly significant as pertains to vaccine effectiveness in terms of individual outcomes. That is to say, there is a measurably higher risk in that window (this is what you find from the UK stats) but when amortized over the 6 month duration of effectiveness it's not really a big deal. (And the Gato post data didn't even find that result, although that could be because they only published part of the data for the worry window). At some point I did the math on the UK data and found IIRC about 2x rate increase over that short time window. For a person getting the shot it's pretty insignificant.
HOWEVER, it's not necessarily insignificant from the standpoint of an epidemic. If that effect causes a brief but substantial increase in infection rate among a significant group of people, it's basically increasing R0 in the short term, leading to spectacular effects in the case rates exploding during rollouts. Because that small rate increase has an exponential effect on case rate. And that in turn can feed back to overall greater harm. And people have also noticed this effect (leading to "the vaccine is the disease" which sometimes goes full tin foil).
The problem is that teasing these two concepts apart is harder to explain and model, so you get some shoddy analysis, leading to vax skeptics being "right" for the wrong reasons (two wrong arguments conflating effects that really are harmful).
I'm a subscriber so I have the israel post in email form if you are interested I can email it to you.
Your explanation was less mathy but maybe still more complicated than needed - if I understand it, his error was adding two fractions with different denominators. Even these days, this is apparently taught in 6th or 7th grade math class.
Yes, in mine bullets 1 and 2 are self-redundant. But they're both my internal "proofs" that I use so I don't have to actually inspect the math when someone says a change to an IRL denominator can change a per-x rate, haha
Thank You for this important post! Didn't Pfizer mention in their own first trial studies of their genetically modifying cocncoction, for the pregnant women to not even get close to the injected?? It is called shedding. I knew personally only one pregnant woman, and she lost her baby after being exposed to all injected grandparents! Here from a TOP lawyer about catastrophic consequences of the injections among the military personnel:
https://www.brighteon.com/4dc4855b-b650-4c66-9c77-2d9a0273ed94
Military deaths and AIDS vaccines -- Muertes militares y vacunas de SIDA
Yes the Pfizer trial protocol pdf includes stark language that “exposure” (including second-hand) during pregnancy must be reported - https://cdn.pfizer.com/pfizercom/2020-11/C4591001_Clinical_Protocol_Nov2020.pdf
But the perfunctory nature of the protocol document should be kept in mind. It’s very important from a CYA standpoint that the text of the document is legally impeccable. It’s totally irrelevant as we have seen whether the text is actually practicable at scale or actually carried out by the trial vendors.