Why is it that the people with three doses of vaccine are always the second-highest in the death count? Is it because older people who have had one shot tend to get three? Like all or nothing?
I’m getting worried that I should have stuck with two.
1 Like
The 80s are a bad age too. Once you get into your 90s, you are good.
Joke before someone starts explaining base rates again
4 Likes
Daily numbers are not representative and are not baserate adjusted. People who refresh their vaccination are less likely to die or suffer serious infections.
I can post the table of current numbers when I get home.
3 Likes
I was hoping that one of you science people could explain this phenomenon.
Same question was asked before. Old people with boosters have much better chances than their peers.
1 Like
Most of us will be refreshing our infections soon enough
Well, look, you can argue about their effectiveness all you like in ordinary use, but if you’re talking about quarantine then yes, it is supposed to be cast-iron all the way. But I think everyone knows that “quarantine” really isn’t any such thing. Just like the masks, it’s a magic ritual that serves no purpose.
As slawa said, it’s just a statistical artifact. There is almost no information being conveyed in these tables because it’s impossible to correct for the many confounding factors involved - the information that you would need to do so just does not exist. Although slawa believes you can adjust for “base rate”, you really can’t. It’s not as simple as just knowing who has had 1/2/3 doses in each age group in the general population because the people on this spreadsheet are not a representative sample of the general population.
Your all-cause risk of death is predicated largely upon all of the ordinary risk factors that you had in 2019, and is modified to no remarkable degree either by COVID itself or by vaccinations. If you want to worry about dying, worry about your risk of heart disease and cancer, because 80% certain one of those will get you before anything else does.
4 Likes
Once again, all-or-nothing. Either it solves the problem entirely, or ir “serves no purpose.”
Guy
Don’t forget traffic! That one can get you too.
Guy
1 Like
Cancer, cardiovascular disease, pneumonia and diabetes account for 72% of all deaths in Presley’s age range. Accidents (of which the majority are probably traffic accidents) are 5%. General respiratory disease (excluding pneumonia) is also about 5%. It’s impossible to know if COVID is modifying that last figure - we won’t know until the end of the year. But yeah, to a first approximation, it looks like you could have more impact on your risk of dying by being extra-careful crossing the street than you could by getting multiple vaccinations.
5 Likes
The previous table I linked is not that great, as it was not divided in age brackets. One can clearly see in the following that older are much more at risk and have good benefits from vaccines.
Deaths per 100k by age by vaccination (2022 until July 6)
| 0 | 1 | 2 | 3 | |
| 0-4 | 1.07 | - | - | - |
| 5-11 | 2.89 | 0 | - | - |
| 12-17 | 3.08 | 0.74 | 0.27 | - |
| 18-29 | 10.81 | 0.9 | 0.29 | 0.1 |
| 30-49 | 54.98 | 9.09 | 3.45 | 1.46 |
| 50-64 | 94.16 | 65.73 | 24.52 | 9.07 |
| 65-74 | 239.31 | 169.71 | 78.11 | 25.78 |
| 75+ | 544.45 | 447.44 | 284.89 | 119.29 |
3 Likes
Yeah, but that still doesn’t explain why people with three shots are dying more than people with one or two.
They are dying at a far reduced rate than their less-vaccinated peers. There’s simply many more of them now, with around 70% of Taiwan’s population having received three shots.
Guy
5 Likes
You’d have to find and check the numbers, but I think it’s like you said before - that the proportions of people who’ve had one or two shots but not three are quite small at this point.
3 Likes
I’m giving this another go, because my previous post was clearly not understood, and @slawa’s table strikes me as so misleading it really shouldn’t be here. I believe the error boils down to incorrectly identifying the population being observed: the denominator in the “base rate correction” is wrong.
For the sake of illustration, let’s take one extremely specific group from the death spreadsheet: people in their 70s who had cancer. Because the MOHW publishes data on cancer, we can make a finger-in-the-air guess at background mortality from this cause in this agegroup. There were 1,450,321 people in this age bracket in 2019, of whom 34,292 died, and one third of those (11652) of cancer. That’s a 12-month all-cause risk of 0.024.
It’s important to grasp the risk stratification that occurs in even in a narrow age range like this. It should be obvious that one cannot die of cancer if one doesn’t have cancer (or heart disease, or whatever). A 70-year-old who definitely doesn’t have cancer has a far, far lower risk of dying of cancer in the next 12 months than someone who has been positively diagnosed, and therefore also a lower risk than the average (he could, of course, develop cancer in that timeframe). A fit, healthy 70-year-old probably has an all-cause risk comparable to the average 50-year-old. On the other hand, a 70-year-old with Stage 3 cancer probably has 20% chance of dying in the next 12 months (given that the prognosis for most cancers is a handful of years).
For the sake of argument, then, let’s say there are 60,000 70somethings living with advanced cancer; a 20% risk yields the expected 11652 deaths a year, or thereabouts. It doesn’t actually matter, for the purposes of calculating relative risks, whether ‘60,000’ is correct or not - it’ll all come out in the wash. But the calculation is clearer if we do it longhand, and we need this number to make a guess at absolute risk.
Of those 60,000 people, 48,000 are (probably) vaccinated, and 12,000 don’t have even one dose.
According to the MOHW spreadsheet, 191 70somethings died with COVID and cancer from 26-May to 26-June. 97 were vaxed, 94 not. But we would have expected 971(ish) in one month anyway. So apparently about 20% of dead cancer patients died ‘of or with’ COVID, and the rest died of … just cancer.
We can do a simple Monte Carlo analysis, varying an excess death rate multiplier, to model the fact that we might have more deaths than usual due to COVID (or indeed other things). We might even have fewer deaths.
| excessrate | expected | actual | noncov | cov+vax | cov+nv | noncov+vax | noncov+nv | allvax | allnv | vaxrisk | nvrisk | RR |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0.8 | 971 | 776 | 585 | 97 | 94 | 468 | 117 | 565 | 211 | 0.011770833 | 0.017583333 | 1.49380531 |
| 1 | 971 | 780 | 97 | 94 | 624 | 156 | 721 | 250 | 0.015020833 | 0.020833333 | 1.386962552 | |
| 1.2 | 1165 | 974 | 97 | 94 | 779 | 194 | 876 | 288 | 0.01825 | 0.024 | 1.315068493 | |
| 1.4 | 1359 | 1168 | 97 | 94 | 934 | 233 | 1031 | 327 | 0.021479167 | 0.02725 | 1.268671193 |
In this table, I’ve taken a guess at the absolute number of cancer deaths not associated with COVID based on 2019 numbers, and split them into vaxed and nonvaxed (nv) groups, based on their ratio in the general population (80:20 or thereabouts). The COVID deaths we know about from the MOHW data, and we know which ones were vaxed and unvaxed.
With the four combinations of covid-deaths/noncovid-deaths, and vaxed/unvaxed, we can calculate the totals for ‘all unvaxed people who died of cancer’ and ‘all vaxed people who died of cancer’. Dividing by the correct populations - the number of cancer patients who have been vaxed (or not) - gives us the risk of death for people expected to die of cancer, vaxed and unvaxed. This is essentially equivalent to ‘rate of death’ (deaths per month in this case). @afterspivak was claiming that this rate is radically different for vaxed and unvaxed.
Turns out that 1-2% of vaxed cancer patients die in a COVID month, and 1.7-2.7% of unvaxed. Well: those numbers are undeniably different. The RR is about 30% … which, as a sanity check, is pretty much what officials are claiming for vaccine effectiveness in other countries. However, these numbers are not different by the ridiculous orders-of-magnitude that slawa is claiming.
It doesn’t make much difference to the RR if 2022 has more or fewer people dying than 2019. However, if you fiddle around with the vax rate, that makes a difference: if we assume 85:15%, the RR changes to about 2 (because the unvaccinated risk looks worse). At 75:25%, the vaccine looks basically useless. Remember we don’t actually know what the vax coverage is for cancer patients; we can only take an educated guess.
I haven’t attempted to split vaccination status into 1, 2 or 3 doses - we have such a small number of subjects, and the effect of these vaccines is so variable, both for the individual and across time (waning efficacy), that the result would be dominated by noise that can’t be dialled out.
No doubt the goalposts will be hurriedly moved, and the cry will revert to “every life is precious!”. Well, is it really? My uncle died of cancer. Towards the end, he would not have wanted his life prolonged by whatever weeks or months a COVID vaccine might grant you. He was ready to go. I don’t think all this buggering about with numbers accurately captures the human reality behind complex medical decisions - and it should be noted that a lot of people on the COVID spreadsheet (120 of them) were annotated “DNR”. They were ready to go.
I may well have made some arithmetic errors here, since it’s 3am. But I’m fairly sure they’re not as egregious as the misrepresentations I’m attempting to correct.
6 Likes
2 Likes
Maybe a lot of those 3rd/booster shots are wearing off - over 6 months for quite a few now.