Read on Twitter
Must Read Deep Dive - Calculating Real Infection Rates from AB Studies
I found this incredible data from NYC AB testing
They've been reporting weekly AB testing results since April 11th (2 mil tests)
This graph has significant ramifications for how we understand COVID-19
1
I found this incredible data from NYC AB testing
They've been reporting weekly AB testing results since April 11th (2 mil tests)
This graph has significant ramifications for how we understand COVID-19
1
Too bad we can't see data from a couple of weeks earlier, but we know NYC was at least 61%!
The first thing we see is the rapid seroreversion in April
This confirms that ABs don't last for the majority of the population
Depending on when you test, results can vary widely
2
The first thing we see is the rapid seroreversion in April
This confirms that ABs don't last for the majority of the population
Depending on when you test, results can vary widely
2
But the most significant aspect is that eventually, you get to a "Stable AB%" - For NYC this is 20%
This number tells us how many people have long-lasting ABs
Working backward, we can use other data to try to predict a maximum value for Stable AB% if 100% are infected
3
This number tells us how many people have long-lasting ABs
Working backward, we can use other data to try to predict a maximum value for Stable AB% if 100% are infected
3
I call the maximum "Stable AB Max"
To calculate this we need to know three variables
-% Symptomatic/% Asymptomatic
-% Symptomatic with long-lasting ABs
-% Asymptomatic with long-lasting ABs
I touched on these variables in my last deep dive on testing:
4 https://twitter.com/gummibear737/status/1300828252869664768?s=20
To calculate this we need to know three variables
-% Symptomatic/% Asymptomatic
-% Symptomatic with long-lasting ABs
-% Asymptomatic with long-lasting ABs
I touched on these variables in my last deep dive on testing:
4 https://twitter.com/gummibear737/status/1300828252869664768?s=20
Last week I said Asymptomatic/Mild were 90-95% but to be precise we need to put the Mild with the symptomatic cases because that how studies on this have been performed
The 95% asymp levels seen in prisons, factories and plants don't account for this
5 https://twitter.com/gummibear737/status/1300828296125599744?s=20
The 95% asymp levels seen in prisons, factories and plants don't account for this
5 https://twitter.com/gummibear737/status/1300828296125599744?s=20
So what do the scientists tell us?
The study with the largest detected % of pure asymptomatics was 80%
Why use the largest %?
Cause seroreversion happens fast - this study probably got lucky and tested on a perfect day
@BallouxFrancois says 70-80%
6 https://www.gov.uk/government/publications/vivaldi-1-coronavirus-covid-19-care-homes-study-report/vivaldi-1-covid-19-care-homes-study-report
The study with the largest detected % of pure asymptomatics was 80%
Why use the largest %?
Cause seroreversion happens fast - this study probably got lucky and tested on a perfect day
@BallouxFrancois says 70-80%
6 https://www.gov.uk/government/publications/vivaldi-1-coronavirus-covid-19-care-homes-study-report/vivaldi-1-covid-19-care-homes-study-report
20% Symp/80% Asymp
What is the percentage of symptomatic cases that maintain long-lasting Antibodies?
It's between 88-92% according to all the AB studies I've seen
I'll use a conservative estimate of 92%
7
What is the percentage of symptomatic cases that maintain long-lasting Antibodies?
It's between 88-92% according to all the AB studies I've seen
I'll use a conservative estimate of 92%
7
Finally, the hardest variable is what % of asymptomatic cases retain long term ABs?
We know it's 80-20 on infected
We also know that the lowest % of asymps ever found in an AB study was 27%
So, infected asymps start 4 times larger than symps, but are only 1/4 in Stable AB%
8
We know it's 80-20 on infected
We also know that the lowest % of asymps ever found in an AB study was 27%
So, infected asymps start 4 times larger than symps, but are only 1/4 in Stable AB%
8
A little bit of math gives us 8.5%
If we run the number we see that the Stable AB Max is 25.2%
9
If we run the number we see that the Stable AB Max is 25.2%
9
Finally:
Divide NYC's Stable AB% by the Stable AB Max
20% / 25.2%
NYC: 79.3% infected
Added a few more cities for comparison (assuming that those are correct Stable AB% for each)
You can argue Stockholm is too low. I chose conservative estimates
10
Divide NYC's Stable AB% by the Stable AB Max
20% / 25.2%
NYC: 79.3% infected
Added a few more cities for comparison (assuming that those are correct Stable AB% for each)
You can argue Stockholm is too low. I chose conservative estimates
10
Here I run the model with 70-30 and 75-25 splits to see how the numbers adjust
I believe 80-20 is the correct split
NYC is definitely at herd immunity!
11
I believe 80-20 is the correct split
NYC is definitely at herd immunity!
11
Side Note: This virus will infect just about everybody
By infected I mean a productive infection that releases at least trace amounts of RNA for detection by PCR
We know this because we've seen prisons with 88% PCR positive inmates
12 https://www.reuters.com/article/us-health-coronavirus-prisons-testing-in-idUSKCN2270RX
By infected I mean a productive infection that releases at least trace amounts of RNA for detection by PCR
We know this because we've seen prisons with 88% PCR positive inmates
12 https://www.reuters.com/article/us-health-coronavirus-prisons-testing-in-idUSKCN2270RX
So the question is: how can Wuhan have only 12% infected?
If you lockdown after the hospitals begin to fill it's already too late
Seems impossible/illogical -only 12% infected
They should have at least 80%
How do you get to 80% infected based on 3% Stable AB%?
97% Asymps
13
If you lockdown after the hospitals begin to fill it's already too late
Seems impossible/illogical -only 12% infected
They should have at least 80%
How do you get to 80% infected based on 3% Stable AB%?
97% Asymps
13
As I've said before, pre-existing immunity is higher in East Asia/South East Asia/Oceana
There's no other way to explain this discrepancy
Wuhan was not social distancing or wearing masks when the virus hit them
All measures were too late
Thread:
14 https://twitter.com/gummibear737/status/1298246346974154757?s=20
There's no other way to explain this discrepancy
Wuhan was not social distancing or wearing masks when the virus hit them
All measures were too late
Thread:
14 https://twitter.com/gummibear737/status/1298246346974154757?s=20
Why more immunity?
The thread below explains,but replace eating bats with close contact/handling wild animals via wet markets/bush meat
Dr. Wolfgang Wodarg says coronavirus often jump wild animal to human (&back), but they're harmless-we don't notice
15 https://twitter.com/gummibear737/status/1298956319173545986?s=20
The thread below explains,but replace eating bats with close contact/handling wild animals via wet markets/bush meat
Dr. Wolfgang Wodarg says coronavirus often jump wild animal to human (&back), but they're harmless-we don't notice
15 https://twitter.com/gummibear737/status/1298956319173545986?s=20
Getting back on track, what does this new information about real infection rates tell us?
Some cities are at or near herd immunity with as little as 10% detectable via current measures if they are at "Stable AB %"
Outside of major urban areas, it could be as little as 6%
16
Some cities are at or near herd immunity with as little as 10% detectable via current measures if they are at "Stable AB %"
Outside of major urban areas, it could be as little as 6%
16
The IFRs are much lower and in the range of what John Ioannidis had predicted ~.25%
NYC and Bergamo are worst-case scenarios - Factors: health of population, no medical knowledge of disease, older populations not protected
17 https://www.medrxiv.org/content/10.1101/2020.05.13.20101253v3
NYC and Bergamo are worst-case scenarios - Factors: health of population, no medical knowledge of disease, older populations not protected
17 https://www.medrxiv.org/content/10.1101/2020.05.13.20101253v3
Here's the whole graph + Calculations in one frame
NYC was at least 61% infected, and it's clear it was decreasing from a much higher %
Note: this data started coming in at the absolute peak of the pandemic in NYC
From peak it takes about 3 months to reach Stable AB%
18
NYC was at least 61% infected, and it's clear it was decreasing from a much higher %
Note: this data started coming in at the absolute peak of the pandemic in NYC
From peak it takes about 3 months to reach Stable AB%
18
I shared this data with @federicolois who's working on an epidemiological model to predict dynamics
His number of total infected are very similar to mine, and he's doing this only with statistical modeling
His model says lockdowns kill more:
Thread:
19 https://twitter.com/federicolois/status/1298630969356890112?s=20
His number of total infected are very similar to mine, and he's doing this only with statistical modeling
His model says lockdowns kill more:
Thread:
19 https://twitter.com/federicolois/status/1298630969356890112?s=20
Regarding a second wave, I'm still not sure
Some cities are already at herd immunity
Maybe a small one
What is clear is that the best thing ever is for students to get back to school, colleges to return to normality, and young healthy people to get on with their lives
20
Some cities are already at herd immunity
Maybe a small one
What is clear is that the best thing ever is for students to get back to school, colleges to return to normality, and young healthy people to get on with their lives
20
And that's it
Sorry, it was so long!
Hope you enjoyed
If you liked it please RT the first tweet in the thread
21
Sorry, it was so long!
Hope you enjoyed
If you liked it please RT the first tweet in the thread
21
