Comparable statistics/risk

Covid-19 discussion, bring your own statistics
User avatar
science_fox
Fuzzable
Posts: 251
Joined: Mon Nov 11, 2019 1:34 pm
Location: Manchester

Comparable statistics/risk

Post by science_fox » Wed Apr 07, 2021 6:45 pm

So the current best guess (it's rare = tricky to be accurate) blood clot risk of dying (less severe possibly more frequent) from the AZ vaccine is 1:100000-250000 depending on age.

That's hard risk to quantify anyone know of some suitable comparisons?

It's much more likely than winning the lottery, but much less likely than winning something on the lottery
It's 1 person in a small/medium? town?

A lot less likely (surprisingly a lot) than dying in a car accident in a given year
A quick google (I might as well do my own research* while I'm here: Makes the vaccine not in the top 18 likely causes of death in the US population, other populations might vary: https://www.livescience.com/3780-odds-dying.html



* not peer reviewed research, research just enough for an internet forum post.

User avatar
Sciolus
Snowbonk
Posts: 556
Joined: Mon Nov 11, 2019 6:42 pm

Re: Comparable statistics/risk

Post by Sciolus » Wed Apr 07, 2021 7:26 pm

Micromort

The blood clot risk of death seems to be about 1 micromort across all age groups. I can't immediately find a figure for under-30s.

User avatar
shpalman
Stummy Beige
Posts: 4031
Joined: Mon Nov 11, 2019 12:53 pm
Location: One step beyond

Re: Comparable statistics/risk

Post by shpalman » Wed Apr 07, 2021 7:36 pm

molto tricky

User avatar
bob sterman
Snowbonk
Posts: 431
Joined: Mon Nov 11, 2019 10:25 pm

Re: Comparable statistics/risk

Post by bob sterman » Wed Apr 07, 2021 7:47 pm

shpalman wrote:
Wed Apr 07, 2021 7:36 pm
https://www.theguardian.com/society/202 ... -regulator has a graphic in it.
The Guardian has only included the graphic for the "Low Exposure Risk" scenario. Things swing strongly in favour of the vaccine when COVID prevalence is much higher...

Image

User avatar
Sciolus
Snowbonk
Posts: 556
Joined: Mon Nov 11, 2019 6:42 pm

Re: Comparable statistics/risk

Post by Sciolus » Wed Apr 07, 2021 7:59 pm

Sciolus wrote:
Wed Apr 07, 2021 7:26 pm
Micromort

The blood clot risk of death seems to be about 1 micromort across all age groups. I can't immediately find a figure for under-30s.
That graphic suggests about 10 micromorts for 20-29-year-olds, but seems to be a bit higher ovrall than other figures I've seen. Not that it matters much given the uncertainty and the fact that it's clearly tiny.

User avatar
bob sterman
Snowbonk
Posts: 431
Joined: Mon Nov 11, 2019 10:25 pm

Re: Comparable statistics/risk

Post by bob sterman » Wed Apr 07, 2021 8:22 pm

Sciolus wrote:
Wed Apr 07, 2021 7:59 pm
That graphic suggests about 10 micromorts for 20-29-year-olds, but seems to be a bit higher ovrall than other figures I've seen. Not that it matters much given the uncertainty and the fact that it's clearly tiny.
It's labelled "serious harms" rather than deaths - I think the fatality rate for the clotting events was 25%. So would that take it down to about 2.5 micromorts?

User avatar
shpalman
Stummy Beige
Posts: 4031
Joined: Mon Nov 11, 2019 12:53 pm
Location: One step beyond

Re: Comparable statistics/risk

Post by shpalman » Wed Apr 07, 2021 8:31 pm

bob sterman wrote:
Wed Apr 07, 2021 7:47 pm
shpalman wrote:
Wed Apr 07, 2021 7:36 pm
https://www.theguardian.com/society/202 ... -regulator has a graphic in it.
The Guardian has only included the graphic for the "Low Exposure Risk" scenario. Things swing strongly in favour of the vaccine when COVID prevalence is much higher...

Image
As it says, that's based on a covid prevalence similar to the peak of the UK's second wave. It's much lower than that now, and falling. The Guardian's graph is based on a level similar to the UK in March.
molto tricky

User avatar
jimbob
After Pie
Posts: 2354
Joined: Mon Nov 11, 2019 4:04 pm
Location: High Peak/Manchester

Re: Comparable statistics/risk

Post by jimbob » Wed Apr 07, 2021 8:42 pm

Meanwhile simply living in the UK from March 2020 to March 2021 has a 2000 micromort risk from COVID-19
Have you considered stupidity as an explanation

User avatar
shpalman
Stummy Beige
Posts: 4031
Joined: Mon Nov 11, 2019 12:53 pm
Location: One step beyond

Re: Comparable statistics/risk

Post by shpalman » Wed Apr 07, 2021 8:45 pm

Well, in March 2021 either you were alive or you weren't. This isn't about whether it would have been worth having the vaccine a year ago or even last month. It's about whether it's worth having it now given how low the covids are now but they'll probably go up when we unlock but maybe not that much if lots of people have been vaccinated.
molto tricky

User avatar
sTeamTraen
After Pie
Posts: 1749
Joined: Mon Nov 11, 2019 4:24 pm
Location: Palma de Mallorca, Spain

Re: Comparable statistics/risk

Post by sTeamTraen » Wed Apr 07, 2021 9:53 pm

bob sterman wrote:
Wed Apr 07, 2021 7:47 pm
The Guardian has only included the graphic for the "Low Exposure Risk" scenario. Things swing strongly in favour of the vaccine when COVID prevalence is much higher...
That was a really bad choice by the Guardian. To the average reader in a hurry, it suggests that the vaccine is slightly more dangerous than COVID to someone under 30.
Something something hammer something something nail

User avatar
shpalman
Stummy Beige
Posts: 4031
Joined: Mon Nov 11, 2019 12:53 pm
Location: One step beyond

Re: Comparable statistics/risk

Post by shpalman » Thu Apr 08, 2021 6:25 am

sTeamTraen wrote:
Wed Apr 07, 2021 9:53 pm
bob sterman wrote:
Wed Apr 07, 2021 7:47 pm
The Guardian has only included the graphic for the "Low Exposure Risk" scenario. Things swing strongly in favour of the vaccine when COVID prevalence is much higher...
That was a really bad choice by the Guardian. To the average reader in a hurry, it suggests that the vaccine is slightly more dangerous than COVID to someone under 30.
Isn't that exactly why the MHRA recommends a different vaccine for those under 30?
molto tricky

User avatar
Bird on a Fire
Light of Blast
Posts: 5561
Joined: Fri Oct 11, 2019 5:05 pm
Location: nadir of brie

Re: Comparable statistics/risk

Post by Bird on a Fire » Thu Apr 08, 2021 8:20 am

shpalman wrote:
Thu Apr 08, 2021 6:25 am
sTeamTraen wrote:
Wed Apr 07, 2021 9:53 pm
bob sterman wrote:
Wed Apr 07, 2021 7:47 pm
The Guardian has only included the graphic for the "Low Exposure Risk" scenario. Things swing strongly in favour of the vaccine when COVID prevalence is much higher...
That was a really bad choice by the Guardian. To the average reader in a hurry, it suggests that the vaccine is slightly more dangerous than COVID to someone under 30.
Isn't that exactly why the MHRA recommends a different vaccine for those under 30?
Yes.
He has the grace of a swan, the wisdom of an owl, and the eye of an eagle—ladies and gentlemen, this man is for the birds!

User avatar
science_fox
Fuzzable
Posts: 251
Joined: Mon Nov 11, 2019 1:34 pm
Location: Manchester

Re: Comparable statistics/risk

Post by science_fox » Thu Apr 08, 2021 8:22 am

https://twitter.com/ReicherStephen/stat ... 3540818947 I'm not sure these are comparable figures though?
If you had the AstraZeneca vaccine today, it was possibly the safest thing you did.
It was safer than having a bath (30 people drown in baths each year)
It was safer than walking down the stairs (1,000 die falling down the stairs each year)
It was safer than eating breakfast...(over 200 people choke on food each year).
It was certainly safer than driving or biking to the shops...
Who seems to be a Scottish prof, but possibly not a statistician. Can you meaningly compare 1 off events with annual figures?

User avatar
bob sterman
Snowbonk
Posts: 431
Joined: Mon Nov 11, 2019 10:25 pm

Re: Comparable statistics/risk

Post by bob sterman » Thu Apr 08, 2021 8:32 am

shpalman wrote:
Thu Apr 08, 2021 6:25 am
sTeamTraen wrote:
Wed Apr 07, 2021 9:53 pm
bob sterman wrote:
Wed Apr 07, 2021 7:47 pm
The Guardian has only included the graphic for the "Low Exposure Risk" scenario. Things swing strongly in favour of the vaccine when COVID prevalence is much higher...
That was a really bad choice by the Guardian. To the average reader in a hurry, it suggests that the vaccine is slightly more dangerous than COVID to someone under 30.
Isn't that exactly why the MHRA recommends a different vaccine for those under 30?
Yes - in the current environment where exposure risk is low.

However, perhaps the "average reader in a hurry" that
sTeamTraen refers to might not appreciate this point?

Probably would have been better to show all three graphs. As If we get a big new wave with a variant that the AZ vaccine can neutralise, there may be a strong rationale for getting under 30s to accept the AZ vaccine then.

User avatar
shpalman
Stummy Beige
Posts: 4031
Joined: Mon Nov 11, 2019 12:53 pm
Location: One step beyond

Re: Comparable statistics/risk

Post by shpalman » Thu Apr 08, 2021 8:54 am

The BBC helpfully explains that 10 deaths in 10 million is 1 in a million after giving a 4 per million chance of blood clots from the AstraZeneca vaccine.

The last figure given for the incidence per 100,000 per week for the UK is 39, so that's 390 per million per week (for England it's the same).

However, for England we can also see the "Cases by specimen date age demographics" where we see that there was recently a bump in cases in the 10-14 and 15-19 ranges (discussed elsewhere on here) and also slightly in the 30-45 range, maybe. Vaccines certainly seemed to have worked to obliterate cases in the older age groups, because those were the last to stop being infected* in previous waves. The rolling rate in the 0-59 age range in England is 46.7 (per 100,000 per week).

It's not particularly easy to estimate the CFR just by looking at Deaths in England because there are so few in the younger age groups and the heat map doesn't obviously let you change the z scale (and I can't be bothered to download the data and extract it again).

But you can see the table of data for deaths in the 0-59 range, and the most recent figure (2nd of April) is 27 for the rolling sum of the past 7 days. The rolling rate is given as 0.1 (per week per 100,000) but that's just their limited precision: I make it 0.066 (for 27 deaths out of a population of 41 million in the 0-59 age range in England). That's 0.66 in a million.

However, it's debatable what time period that probability applies to, because it's just based on who died over the past 7 days having tested positive for covid within 28 days before that.

Cases are coming down so deaths should be coming down too but with a lag. If I compare deaths on the 2nd of April with cases two weeks before that, I get almost exactly 1% for the CFR in the 0-59 age range. Someone who can be bothered to extract the JSONs can work this out for the under-30 age range in a similar way.

But based on these figures for the 0-59 age group in England, about 467 in a million will test positive for covid in the next week and 4-5 in a million will end up dying from it.

There have been 79 blood clot cases out of 20m vaccinations, 19 of whom died.

So about 1 in a million will die as a result of the AstraZeneca vaccine.

However, as the Guardian's graph points out, because it was created by people who are not idiots and who have the actual data, since the chance of getting a positive covid test and then dying of covid decreases a lot as you get younger, while the risk of an adverse reaction to the vaccine increases, if the overall incidence of covid in the population is low enough there is more risk than benefit for the AstraZeneca vaccine for young people.

It does kind of assume that there won't be another wave, and/or that the AstraZeneca vaccine wouldn't be efficacious against whatever variant is waving by then anyway.
molto tricky

User avatar
shpalman
Stummy Beige
Posts: 4031
Joined: Mon Nov 11, 2019 12:53 pm
Location: One step beyond

Re: Comparable statistics/risk

Post by shpalman » Thu Apr 08, 2021 9:05 am

science_fox wrote:
Thu Apr 08, 2021 8:22 am
https://twitter.com/ReicherStephen/stat ... 3540818947 I'm not sure these are comparable figures though?
If you had the AstraZeneca vaccine today, it was possibly the safest thing you did.
It was safer than having a bath (30 people drown in baths each year)
It was safer than walking down the stairs (1,000 die falling down the stairs each year)
It was safer than eating breakfast...(over 200 people choke on food each year).
It was certainly safer than driving or biking to the shops...
Who seems to be a Scottish prof, but possibly not a statistician. Can you meaningly compare 1 off events with annual figures?
I'm well aware that driving to Milan and back to get my vaccine was far more risky than the vaccine itself.

But I didn't skip breakfast, avoid stairs, or change the amount of baths I took because of it, so the vaccine did (very very slightly) increase my risk of death.

Spoiler: I didn't die.

* Lockdown means it's been over a year since I've made that trip whereas I used to do it once or twice a week, so overall covid has probably extended my life.
molto tricky

User avatar
shpalman
Stummy Beige
Posts: 4031
Joined: Mon Nov 11, 2019 12:53 pm
Location: One step beyond

Re: Comparable statistics/risk

Post by shpalman » Thu Apr 08, 2021 9:09 am

bob sterman wrote:
Thu Apr 08, 2021 8:32 am
shpalman wrote:
Thu Apr 08, 2021 6:25 am
sTeamTraen wrote:
Wed Apr 07, 2021 9:53 pm

That was a really bad choice by the Guardian. To the average reader in a hurry, it suggests that the vaccine is slightly more dangerous than COVID to someone under 30.
Isn't that exactly why the MHRA recommends a different vaccine for those under 30?
Yes - in the current environment where exposure risk is low.

However, perhaps the "average reader in a hurry" that
sTeamTraen refers to might not appreciate this point?

Probably would have been better to show all three graphs. As If we get a big new wave with a variant that the AZ vaccine can neutralise, there may be a strong rationale for getting under 30s to accept the AZ vaccine then.
Why would we get a big new wave of a variant which a substantial fraction of the population has been vaccinated against?
molto tricky

User avatar
Brightonian
Catbabel
Posts: 633
Joined: Mon Nov 11, 2019 3:16 pm
Location: Usually UK, often France and Ireland

Re: Comparable statistics/risk

Post by Brightonian » Thu Apr 08, 2021 9:09 am

science_fox wrote:
Thu Apr 08, 2021 8:22 am
https://twitter.com/ReicherStephen/stat ... 3540818947 I'm not sure these are comparable figures though?
If you had the AstraZeneca vaccine today, it was possibly the safest thing you did.
It was safer than having a bath (30 people drown in baths each year)
It was safer than walking down the stairs (1,000 die falling down the stairs each year)
It was safer than eating breakfast...(over 200 people choke on food each year).
It was certainly safer than driving or biking to the shops...
Who seems to be a Scottish prof, but possibly not a statistician. Can you meaningly compare 1 off events with annual figures?
Assuming it's a one-off event. Maybe we'll have to have an annual covid jab if we get variants that vary sufficiently.

User avatar
shpalman
Stummy Beige
Posts: 4031
Joined: Mon Nov 11, 2019 12:53 pm
Location: One step beyond

Re: Comparable statistics/risk

Post by shpalman » Thu Apr 08, 2021 9:17 am

How many AZ doses has the UK given to anyone under 30 anyway, and why? When would your age-based rollout have gotten to them?
molto tricky

User avatar
bob sterman
Snowbonk
Posts: 431
Joined: Mon Nov 11, 2019 10:25 pm

Re: Comparable statistics/risk

Post by bob sterman » Thu Apr 08, 2021 9:43 am

shpalman wrote:
Thu Apr 08, 2021 9:17 am
How many AZ doses has the UK given to anyone under 30 anyway, and why? When would your age-based rollout have gotten to them?
They wouldn't have got it through age-based roll-out yet.

Not sure of the number but there were a lot <30 vaccinated as part of Group 6. People with various clinical conditions putting them at risk - not the "clinically extremely vulnerable" - a much broader list than that including obesity, diabetes. Also unpaid carers, and those receiving carers allowance.

Many of the under 30s working in the NHS tended to get the Pfizer jab - but I think a lot of people working in social care, and people working in care homes got AZ. And many are under 30.

Personally I know lots of people in their 30s and 40s who've had the AZ jab and none got it through age-based roll-out. So there will be plenty of people under 30 in the same boat.

User avatar
Woodchopper
Stummy Beige
Posts: 3030
Joined: Sat Oct 12, 2019 9:05 am

Re: Comparable statistics/risk

Post by Woodchopper » Thu Apr 08, 2021 9:58 am

science_fox wrote:
Thu Apr 08, 2021 8:22 am
https://twitter.com/ReicherStephen/stat ... 3540818947 I'm not sure these are comparable figures though?
If you had the AstraZeneca vaccine today, it was possibly the safest thing you did.
It was safer than having a bath (30 people drown in baths each year)
It was safer than walking down the stairs (1,000 die falling down the stairs each year)
It was safer than eating breakfast...(over 200 people choke on food each year).
It was certainly safer than driving or biking to the shops...
Who seems to be a Scottish prof, but possibly not a statistician. Can you meaningly compare 1 off events with annual figures?
There is also the problem of presenting population wide risks and comparing them to risks associated with a specific sub-group (women aged below 60 according to the EDM).

Similarly, the above accidents are do not occur randomly among the UK population. Certain sub-groups are far more likely to choke on food, fall down the stairs or drown in the bath etc. For example, dementia must be a high risk.

So a much better comparison would be the rate of those fatal accidents among women under 60 with the risks associated with the vaccine.

Chris Preston
Clardic Fug
Posts: 230
Joined: Tue Nov 12, 2019 8:05 am

Re: Comparable statistics/risk

Post by Chris Preston » Thu Apr 08, 2021 10:04 am

One of our medical epidemiologists announced on television this evening that the risk was lower than getting struck by lightening.

For what it is worth, lightening is responsible for 100 injuries and between 5 and 10 deaths every year in Australia. 80% of injuries result from using landlines during thunderstorms.
Here grows much rhubarb.

User avatar
sTeamTraen
After Pie
Posts: 1749
Joined: Mon Nov 11, 2019 4:24 pm
Location: Palma de Mallorca, Spain

Re: Comparable statistics/risk

Post by sTeamTraen » Sat Apr 10, 2021 12:37 pm

Chris Preston wrote:
Thu Apr 08, 2021 10:04 am
One of our medical epidemiologists announced on television this evening that the risk was lower than getting struck by lightening.

For what it is worth, lightening is responsible for 100 injuries and between 5 and 10 deaths every year in Australia. 80% of injuries result from using landlines during thunderstorms.
"Getting struck by lightning" is often portrayed as the canonical random event, but like getting run over by a bus, it's not entirely independent of someone's behaviour. Getting hit by a meteorite, now that's more like it.
Something something hammer something something nail

User avatar
dyqik
Stummy Beige
Posts: 3595
Joined: Wed Sep 25, 2019 4:19 pm
Location: Masshole
Contact:

Re: Comparable statistics/risk

Post by dyqik » Sat Apr 10, 2021 1:31 pm

sTeamTraen wrote:
Sat Apr 10, 2021 12:37 pm

"Getting struck by lightning" is often portrayed as the canonical random event, but like getting run over by a bus, it's not entirely independent of someone's behaviour. Getting hit by a meteorite, now that's more like it.
Well...

Someone living at the bottom of a well is a lot less likely to get hit by a meteorite and more likely to die of radon exposure than the average.

They're also very unlikely to be hit by lightning.

User avatar
Martin Y
After Pie
Posts: 1886
Joined: Mon Nov 11, 2019 1:08 pm

Re: Comparable statistics/risk

Post by Martin Y » Sat Apr 10, 2021 3:19 pm

If the link was real and the risk of death from a clot was 1 per million, then continuing to vaccinate the adult population would cause a few tens of deaths. (Stopping vaccinating at all would of course lead to more deaths.) Delaying vaccination until alternatives are available would presumably cause some deaths, but I wonder where the crossover point is with the suspected clotting risk.

Post Reply