Share this page!

Atomsk's Sanakan Profile picture
Twitter logo
23 Dec, 43 tweets, 31 min read
1/X
Those behind the Great Barrington Declaration mention herd immunity as a way to address COVID-19.

So I'll discuss it. After all, noting herd immunity (in response to vaccine deniers) is 1 main reason I started on Twitter.



gbdeclaration.org Image
2/X
Suppose u want to know how many people would die from COVID-19 under *baseline conditions*.

So basically: treat COVID-19 like another typical disease, with business-as-usual and acting the same as this time last year without the pandemic.

3/X
Re: "how many people would die from COVID-19 under *baseline conditions*"

One can figure that out using:
- the number of people who would get infected
- how many of those infected people die of COVID-19

A separate thread on the latter point:
4/X
The virus infects more people if the virus is more contagious, meaning it has a larger R0.

R0 of 2 implies each infected person infects 2 other people on average, at baseline, before they stop being infectious.
For R0 of 3, they infect 3 others.
Etc

mdpi.com/2076-393X/8/2/… Image
5/X
Eventually, so many people become non-infectious + immune to infection, that the virus struggles to find non-immune people to infect and use to infect other non-immune people.

That's the idea behind herd immunity.

medium.com/@silentn2040/t… Image
6/X
The 'herd immunity threshold' (HIT) is the proportion of people who need to be immune to infection, in order for 'infections per unit time' to stop increasing (i.e. keep R under 1), at baseline.

At HIT, the outbreak is on its way to dying out.

7/X
The classic calculation for HIT is:
1 - (1 / R0)

So an R0 of 3 implies:
HIT = 1 - (1/3) = 67%

A larger R0 means a larger the HIT.

In other words: a more contagious virus means more people need to be immune to infection for the outbreak to die out.

sciencedirect.com/science/articl… Image
8/X
Different places have different baseline conditions, + thus different values for R0 and HIT.
journalofinfection.com/article/S0163-…

A typical R0 for a western country is ~2.5 or more, implying at HIT of >= 60%. Higher than seasonal flu.
link.springer.com/article/10.118…

thelancet.com/action/showPdf… Image
9/x
People still get infected after HIT is reached, but not enough to replace the people who become immune after they recover from infection (since R is now less than 1).

This implies "overshoot": the final percentage of people infected is more than HIT

Image
10/X
So for the question from part 2/X:

How many people would die from COVID-19 under *baseline conditions*?

Ferguson et al. answered this in March, with a HIT of ~58%, and IFR of 0.9% for Great Britain (~0.8% for the USA):



web.archive.org/web/2020042101… Image
11/X
So how does this play out in reality?

Well, a respiratory virus like SARS-CoV-2 that spreads by droplets + aerosols, takes longer to infect a given proportion of people in:
- a larger population
- a population spread over a wider geographical area

12/X
Since SARS-CoV-2 is very contagious (high R0 and therefore high HIT), SARS-CoV-2 then quickly infects a large proportion of people in smaller populations and/or populations covering small areas, before behavior changes and interventions limit spread

Image
13/X
SARS-CoV-2 also infects a higher proportion of people in areas that remain closer to the baseline conditions of R0 (i.e. not much infection-limiting behavior changes and/or public health interventions like mask-wearing).



web.archive.org/web/2020110203… Image
14/X
Larger populations see people dying of COVID-19, and respond with additional behavior changes + public health interventions that push them further from the baseline conditions of R0.

That limits the spread of SARS-CoV-2 and limits COVID-19 deaths.

Image
15/X
So larger populations and/or populations spread over larger geographic areas, end up with a lower proportion of people infected, even though HIT is high.

Some people see those lower proportion of infected people, and incorrectly infer HIT is low

Image
16/X
But remember, it wasn't herd immunity that limited infections into larger populations and regions; those regions didn't reach HIT.

Instead, it was behavior changes and/or public health interventions that limited infections



ncbi.nlm.nih.gov/pmc/articles/P… Image
17/X
I'll make a claim some people may find controversial:

Claiming HIT is very low (ex: ~10% - ~20%) is *dangerous and obviously incorrect.*
In fact, it may be the most dangerous idea to emerge during the COVID-19 pandemic.



Image
18/X
Saying we reached a low HIT tells us we no longer need to go beyond *baseline conditions* to prevent infections/day from increasing; i.e. no additional:

- mask-wearing
- avoiding visiting nursing homes + large indoor gathers
- vaccinations
etc.

Image
19/X
Some non-experts and experts claim HIT is low.

Examples of non-experts:

medrxiv.org/content/medrxi…
zerohedge.com/political/whit…
realclimatescience.com/2020/06/no-imm…
off-guardian.org/2020/07/07/sec…
medium.com/@jrfinkel/covi…



judithcurry.com/2020/05/10/why…

judithcurry.com/2020/06/28/the… Image
20/X
The non-experts have no background in epidemiology, immunology, etc.

So they falsely assume only herd immunity can limit R and thus limit infections per day; i.e. they assume if infections/day and COVID-19 deaths/day decrease, then HIT was reached

archive.is/h96zK#selectio… Image
21/X
But factors other than herd immunity can limit infections and deaths, as covered in part 16/X. So the non-experts are wrong.

You can usually spot these non-experts because they claim Stockholm, Sweden (or New York City, or...) reached HIT.

Yet...
archive.is/CKncm#selectio… Image
22/X
So what about experts who claim HIT is low? Probably the most well-known one is Gabriela M. Gomes (@mgmgomes1).

Unlike the non-experts, her team is aware that HIT is about baseline conditions of R0.



medrxiv.org/content/10.110…

medrxiv.org/content/10.110… Image
23/X
But the expert proponents of low HIT still need to distinguish the effects of HIT, vs. the effects of behavior changes + public health interventions.

Turns out Gomes' team did that incorrectly.



medrxiv.org/content/10.110… Image
24/X
When one better accounts for the effects of behavior changes + public health interventions, HIT is >50% (green), instead of ~10% - ~20% (blue).

Achieving this higher HIT, without a vaccine, would cause more COVID-19 deaths.



medrxiv.org/content/10.110… Image
25/X
Other sources also support a HIT of >50%, or HIT still not being reached at a >40% infection rate.

web.archive.org/web/2020082310…
web.archive.org/web/2020090113…

"A method is presented for estimating the model parameters from real-world data [...]"
journals.plos.org/plosone/articl… Image
26/X
Why think HIT is high?

One reason centers on heterogeneity vs. homogeneity.
Or in layman's terms: differences vs. sameness.

The "HIT = 1 - (1 / R0)" calculation from part 7/X assumes sameness, while low HIT proponents claim large differences.

sciencedirect.com/science/articl… Image
27/X
If "heterogeneity vs. homogeneity" is confusing, think of a sexually-transmitted infection (STI) like HIV.

With STIs, heterogeneity is high (so large differences virus transmission interactions).


cambridge.org/core/services/…

archive.is/qBEuG#selectio… Image
28/X
But SARS-CoV-2, the virus that causes the disease COVID-19, isn't an STI. It's a respiratory virus spread by droplets + aerosols, using behaviors more common to everyone, such as breathing + face-touching

So there's more sameness (i.e. homogeneity)

quantamagazine.org/the-tricky-mat… Image
29/X
Ironically, many non-experts try to lecture me on how heterogeneity (differences) are large for SARS-CoV-2, when they know less about this than me. 🤦‍♂️

Highlights in tweets in part 30/X onwards, in case they try this on you.



Image
30/X
Different T cell responses between people won't give enough heterogeneity to greatly lower HIT, especially since T cells are not primarily involved in limiting infections. They're more about responding after infection.



Image
31/X
In layman's terms: cross-reactivity involves the immune system treating SARS-CoV-2 like another virus the immune system previously responded to, such as another coronavirus.

Cross-reactivity isn't going to drop HIT by a lot



nature.com/articles/s4157… Image
32/X
There are transmission differences, such as medical professionals generating aerosols when they intubate people (i.e. place tube down their throat), placing those professionals at more risk from SARS-CoV-2-containing aerosols.

web.archive.org/web/2020122322… Image
33/X
But these differences occur for other respiratory viruses such as influenza, w/o causing a much lower-than-expected HIT.

Other pertinent differences are likely already included in R0 (see parts 28/X and 31/X).



archive.is/8MiEc#selectio… Image
34/X
There isn't perfect sameness (perfect homogeneity).

But it's homogenous enough for a high HIT + to have "HIT = 1 - (1 / R0)" from part 7/X be a decent approximation, consistent with the high infection rates from 12/X + 13/X



medrxiv.org/content/10.110… Image
35/X
Proponents of low HIT also predicted areas with high infection rates wouldn't have strong second waves.

They were wrong (another example in part 21/X).





archive.is/m8e89#selectio…
Image
36/X
Gomes' low HIT framework predicted places with higher infection fatality rates would have lower infection rates.

That didn't consistently hold up.



mdpi.com/2079-7737/9/6/…
web.archive.org/web/2020083021…



Image
37/X
The observed pattern of infections and COVID-19 deaths better fit one would expect from behavior changes + public health interventions limiting infections, not herd immunity (with a low HIT) limiting infections, as per part 16/X.

ncbi.nlm.nih.gov/pmc/articles/P… Image
38/X
There are other region-specific reasons to think particular regions did not reach a low HIT. I've covered some elsewhere:



medium.com/@silentn2040/t… Image
39/X
So some reasons for thinking the herd immunity threshold is high:

- the biology underlying transmission of respiratory viruses
- high infection rates
- second waves
- higher fatality rates at higher infection rates
etc.

And I've see no good reason to think HIT is very low
40/X
Some folks claim HIT is low, b/c it allows them to downplay how dangerous COVID-19 is + thus avoid policies they dislike (like lockdowns).

Hence why many of the same people who suggest HIT is low, also under-estimated SARS-CoV-2's fatality rate. 🤷‍♂️

Image
41/X
Also, other factors can increase HIT.

For example: people not becoming immune to infection after they're infected. The "HIT = 1 - (1 / R0)" calculation assumes persistent immunity after infection, as per part 26/X.



archive.is/Xjyec#selectio… Image
42/X
Though SARS-CoV-2 is more contagious than seasonal influenza (i.e. higher R0), other factors slow its spread. That gives larger populations more time to use public health interventions + behavior changes to slow its spread, as per parts 11/X to 16/X

who.int/docs/default-s… Image
43/X
Re: "Some folks claim HIT is low, b/c it allows them to downplay how dangerous COVID-19 is + thus avoid policies they dislike (like lockdowns)"

The Great Barrington Declaration is a vehicle for that



cell.com/med/fulltext/S…

Image

• • •

Missing some Tweet in this thread? You can try to force a refresh
　

Keep Current with Atomsk's Sanakan

Atomsk's Sanakan Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!

PDF

Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @AtomsksSanakan

Atomsk's Sanakan Profile picture
22 Dec
1/
Many COVID-19 contrarians, including those behind the Great Barrington Declaration, *still* cite John Ioannidis' inaccurate estimate of SARS-CoV-2's fatality rate.

So let's go over how atrocious Ioannidis' paper is.



web.archive.org/web/2020111809…
2/
Background:

When a virus infects u, your body increases production of proteins known as antibodies, which are usually specific to that virus.

So measuring antibodies lets u estimate who was infected, and from that the infection fatality rate (IFR).

institutefordiseasemodeling.github.io/nCoV-public/an…
3/

Ioannidis uses antibody (a.k.a. seroprevalence) studies to estimate the number of people infected with the virus SARS-CoV-2. He then calculates IFR by dividing the number of COVID-19 deaths by the number of infected people.

Ioannidis does this badly:
medrxiv.org/content/10.110…
Read 39 tweets
Atomsk's Sanakan Profile picture
8 Dec
1/P

Peter C Gøtzsche (@PGtzsche1) wrote the article below

He argues that COVID-19 isn't very lethal, + then draws some political conclusions.

The article is poor.

"Is the infection fatality rate for COVID-19 worse than that for influenza?"
bmj.com/content/371/bm…
2/P

Gøtzsche's basic idea is:
The proportion of SARS-CoV-2-infected people who die of the disease COVID-19 is comparable to that of flu; i.e. the infection fatality rate (IFR) for COVID-19 is not an order of magnitude larger than that of the flu.

So:
bmj.com/content/371/bm…
3/P

Gøtzsche is wrong. Study after study shows that the fatality rate for SARS-CoV-2 is about an order of magnitude larger than that of influenza; COVID-19 is way more dangerous than the flu.

So where does Gøtzsche go wrong?



link.springer.com/article/10.100…
Read 16 tweets
Atomsk's Sanakan Profile picture
4 Dec
The USA's SeroHub is out. It purports to cover SARS-CoV-2 seroprevalence studies in the USA.

It includes the quite bad, non-peer-reviewed Santa Clara study. So I thought I'd include some other USA studies SeroHub now leaves out.



covid19serohub.nih.gov
Re: "some other USA studies SeroHub now leaves out"

Baton Rouge, Louisiana
medrxiv.org/content/10.110…

Ohio
arxiv.org/abs/2011.09033
coronavirus.ohio.gov/static/dashboa…

Washoe County, Nevada
washoecounty.us/outreach/2020/…

4 Utah counties
medrxiv.org/content/10.110…
Re: "some other USA studies SeroHub now leaves out"

Maricopa County, Arizona
maricopa.gov/CivicAlerts.as…
maricopa.gov/5607/COVID-19-…

Orange County, California
medrxiv.org/content/10.110…

Riverside County, California
rivcoph.org/Portals/0/Docu…

Connecticut
amjmed.com/article/S0002-…
Read 5 tweets
Atomsk's Sanakan Profile picture
25 Nov
It's almost 2021.
And I don't forget, even years later. 🙂



For analyzing global warming trends:
ysbl.york.ac.uk/~cowtan/applet…
psl.noaa.gov/cgi-bin/data/t…
climexp.knmi.nl/start.cgi

Judith Curry (@curryja), in 2016:
judithcurry.com/2016/03/06/end…
Someone will have a lot to answer for in 2021...

"I do receive some funding from the fossil fuel industry"
scientificamerican.com/article/climat…

"Politically I am independent, with libertarian leanings."
culturalcognition.net/blog/2014/8/19…



judithcurry.com/2011/10/27/can…
And for those wondering which dataset to use in assessing Curry's claims.



Read 4 tweets
Atomsk's Sanakan Profile picture
20 Nov
PaperOfTheDay

@GidMK @BillHanage

Gangelt, with:
- an IFR of ~0.4%
- 13.6% IgG+ (which they test-corrected to 14.1%)

"Infection fatality rate of SARS-CoV2 in a super-spreading event in Germany"
nature.com/articles/s4146…

However:

finddx.org/covid-19/dx-da…

medrxiv.org/content/10.110… Image
So the test correction based on FINDDx increases the Gangelt IFR by ~4, and/or renders Gangelt less relevant since corrected seroprevalence overlaps with 0%.

There are other cases in which using FINDDx decreases IFR. For example:



medrxiv.org/content/10.110… Image
Based on the above paper's FINDDx-based analysis, some of the worse offenders in terms of sensitivity (i.e. false negatives) at particular times post-infection are:

- 41%: Abbott Architect IgG
- 42%: Euroimmun IgG
- 57%: CTK Biotech IgG/IgM rapid test

Read 5 tweets
Atomsk's Sanakan Profile picture
8 Nov
1/S

Sweden did not lockdown in response to COVID-19. Many politically-motivated COVID-19 contrarians try to support Sweden's policy by making misleading comparisons between Sweden + other countries. This thread will address that.



Image
2/S

On average, it takes a virus less time to spread thru + kill 50% of a population of 10 people vs. 50% of a population of 10,000 people.

So smaller populations have less time for public health changes, behavior changes, etc. to limit their COVID-19 deaths per capita.
3/S

Also, on average, it takes less time for a virus to spread thru a smaller geographical area.

So microstates' combination of smaller geographical area and smaller population make them a poor, apples-to-oranges comparison to Sweden.

ourworldindata.org/coronavirus-da… Image
Read 16 tweets

Did Thread Reader help you today?

Support us! We are indie developers!

This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Too expensive? Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal Become our Patreon

Thank you for your support!

Follow Us on Twitter!

Like this author's thread?

Get emails when @AtomsksSanakan has new unrolls!

Check out other threads from @AtomsksSanakan!

Read all threads by @AtomsksSanakan