Ben M. Profile picture
30 Sep, 44 tweets, 15 min read
🚨 Does #Covid19 Vaccination lead to less COVID-19-Like Illness (CLI)?

A test-negative design study with 41,552 participants looked CLI diagnosis, symptomatic illness, and test positivity by vaccination status.

⬇️ Megathread below... ⬇️ Like & RT 🚨
#Covid #Corona #Coronavirus Image
Let's start from the beginning!

A recent study by Thompson et al., which was published in the @NEJM, funded and features by @CDCgov in its weekly MMWR series, looked at COVID-19 Like Illness (CLI) hospitalizations for the age group 50+. ImageImage
The study authors identified over 40k hospitalizations.

These were not all-cause hospitalizations, but CLI diagnosed hospitalizations, which is a subset of all hospitalizations. Time frame: 1/1/21 - 6/22/21.

The below diagram tries to visualize the study sample (red). Image
They gathered 103,199 hospitalizations from 7 diff. locations, see map.

They used a test-negative study design, means that they included everyone that matched the criteria (50+ & CLI diagnosis). Image
This is not ideal, since it’s still not all-cause hospitalization by vaccine status, but it’s much better than what we’ve seen before.

And it should be representative within it's sample.
It is noteworthy, that they excluded 61,647 hospitalizations, that did not match their study criteria:
1) <50
2) No vaxx records
3) Repeat admissions
4) No covid19 testing
5) Recently Vaxxed (<14d) Image
Excluding 1-4 makes sense.

Excluding 5, dose1 vaccinated <14 days, does not make any sense to me! Are these hospitalizations in connection with the vaccine?
In the first chart, I have included the fifth category, specifically 1,872 CLI hospitalizations that were excluded by the authors.
Notably here, the authors also made a simple math mistake! Unbelievable!! Image
So the main result of their study was:

"The effectiveness of full messenger RNA (mRNA) vaccination (≥14 days after the second dose) was 89% [...] against laboratory-confirmed SARS-CoV-2 infection leading to hospitalization [...]"

OK, so 89% effective. Great, right? Image
I have calculated Covid19 Hosp. Vaccine Efficacy (VE) based on the raw numbers, and I arrive at 91%. That's close, and the reason why it's off by 2% is the additional statistical corrections the authors did (age, testing bias, etc.). Image
Ok, so what’s the problem with this?

The problem remains to be the highly sensitive PCR test. Confirmed covid in this case is defined as CLI & PCR+. Same as in the vaccine studies, see here Pfizers.

They did not tell us the rate of all-cause sickness or CLI! Image
So if we calculate the vaccine efficacy in the same manner as before, taking into account all the CLI admissions including the excluded 1,872 visits (dose1 <14 day vaxxed), I calculate a VE of -13%! ImageImage
Important note:

All following charts are excluding the 1,872 “patients receiving dose-1 <14 days prior to index hospitalization date”, since they were excluded in the final study sample, and no detailed information is available!
So, based on the study sample of 41,552, what if we only look at how many people actually got sick?
If we actually look at how many people were diagnosed by their physician with CLI, we can see that all three groups were diagnosed with about the same rate... Odd right? Image
Here's the same chart, CLI clinical diagnosis, with the COVID-19 RT-PCR test positivity (blue).
We can observe that while CLI diagnosis remains steady; Covid cases are significantly lower in vaccinated. Why is that? More on that topic later below… Image
First let's look at actual Covid-Like-Illness (CLI) symptoms! This is what IMO should be the relevant metric, as this is a sign that a human is potentially sick FROM a respiratory virus.

We observe the same effect, the rate of symptomatic CLI is very similar! Image
Let's see how these people tested for Covid-19 with the PCR test?

Similar picture as before, while partially vaccinated, and vaccinated are slightly more symptomatic, they test less positive for the SARS-CoV-2 virus. Image
The differences between symptomatic CLI could be by chance, or could it be possible, that they are due to the vaccine itself?

Specifically, partially vaccinated had the most symptoms!

Note, that this does not even include the 1,872 excluded dose1/<14day patients!!! Image
In the same manner as the original authors, we can now calculate the Vaccine Efficacy against symptomatic Covid-Like-Illness (CLI).

Huh?! Negative too. Yep, that's right. Because based on the data, the vaccinated groups actually experience more CLI symptoms. Image
Interesting, right? So this means that despite Covid19 vaccination, people appear to get as sick and hospitalized (if not even more!), as before?!

Here are some other aspects that I looked at:
1) Age:
Observation: Older age groups have higher vaxx rate, so these charts are in line with expectations. Image
2) Comorbidities
Did not appear to be a driving factor as the difference was within 2-3 percentage points.
3) Hospitalization, that later lead to ICU admission
Between 16-20% of initially as CLI admitted later went to ICU.
There’s a slight difference in favor of vaxxed. However, this diff is rather small in magnitude. Could be due to confounders/chance, physician bias, exclusion... Image
Assuming these were accurate numbers, it would only yield a 20% VE, which would be modest, and way below CDC’s 50% threshold. Image
4) Rank of CLI diagnosis

How they coded the CLI diagnosis, primary vs non-primary.
We can see that the primary CLI diagnosis decr. by about half from unvaxxed to fully vaxx. Could this also be due to physician bias?
Since we know that symptoms are equal between groups. Image
5) Testing Propensity (Bias) between the groups.
Based on the data provided, I find that unvaccinated hospitalized CLI/50+ patients, are 11% more often tested for Covid19, or vaccinated CLI/50+ patients are 12.5% less often tested. Image
Summary/Interpretation:
1) Rate of CLI admission, diagnosis, symptoms/signs similar between unvaccinated vs vaccinated (possibly even higher in vaccinated) → No sign. difference in covid like illness between vaxx groups.
2) Study sample exclusions lead to distortion of data:
such as 1,872 excluded dose1 <14 day vaxxed.
Such as 1,847 potentially excluded vaccinated CLI’s due to testing bias (-12.5%/11%)
3) Rate of Covid19 test positives are significantly lower, but why? See some explanations below:
Possible Explanations/Open Questions:
So now that we have proven, that vaccinated people do not get significantly less sick with CLI and therefore in general, what might actually be happening?
We do not know for certain, as this would require all-cause hospitalization/death data by vaxx status. No country (afaik) publishes this, specifically including # of performed tests by vaxx status.
CDC has acknowledged to me that they have this data, but they do not publish it and have denied my FOIA request (for now)... Why?

docs.google.com/document/d/1EG…
We can only list out some hypothesis:

A) The vaccines are effective in eradicating the virus, at given VE (90%), BUT this does not lead to less sickness (CLI).
(i) Is similar sickness, then caused by other pathogens?
(ii) If the vaccine is eradicating a virus, but people still get sick at the same rate, what's the purpose of mass vaccinations?
B) Is there a difference in testing? We know everyone in the study sample is tested, but are they using the exact same tests (protocols, CT, # of genes, etc.)?
C) Is illness severity decreasing due to the vaccine?
In the data, we can see a little less ICU admissions in the vaccinated groups. However the decrease is only one fifth of the claimed VE.
..C) Hence other factors, such as chance, physician bias, etc. could be at play here. Also we could see that exclusions and testing propensity distort the data towards unvaccinated.
Additional thoughts:
In my opinion this proves that the roughly same amount of people in each group are hospitalized, regardless of the Covid19 vaccination status, and that PCR+ does not seem to lead to less CLI.
We can observe this in the real world. See here for examples for Germany and the US.

Note, that the CLI diagnosis (orange) is not very much correlated to the total hospitalizations. This is likely due to the same effect as described above. ImageImage
Methodology/Sources:

CDC MMWR: cdc.gov/mmwr/volumes/7…
Study: nejm.org/doi/10.1056/NE…
Calculations: docs.google.com/spreadsheets/d…
Most of the data, used for calculations was extracted from this table, of the study appendix.
nejm.org/doi/suppl/10.1… ImageImage
Thanks to:
@shasafaya for pointing me to this study & working with me on this!
@prof_shahar for comment, feedback and advice!
@tlowdon for pointing me to the US hospitalization dashboard.
Everyone for following, liking, sharing and commenting!

Please let me know what you think!
Please let me know if there are any errors/mistakes in this analysis.

@threadreaderapp unroll please
And if you haven't seen my website yet, check out USMortality.com for latest US all-cause (pop/age corrected) mortality data.

Maybe @CDCgov can let us know why we're seeing huge excess deaths again in the US... Is it really all the unvaxxed?
Please just publish the data! Image

• • •

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

Keep Current with Ben M.

Ben M. 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 @USMortality

20 Oct
Mortality data up to week 40/2021 now available on USMortality.com!
Mortality as z-score chart.

This years summer wave about the same, a bit higher than last year, but it appears to be ending now...
US Mortality by age group.
Read 5 tweets
13 Oct
New update including week 39 (40), ending 10/02/2021 is up on USMortality.com!

#Covid #Corona #Covid19 #coronavirus
Here by age groups!
Note this is all by z-score... You can find absolute numbers on my website as well!
Here's a map of excess mortality...

We can see the move up to the north...
Read 4 tweets
13 Oct
Comparing Excess Deaths and #COVID-19 Deaths (Underlying Cause) Week 1-38, 2020 vs 2021.

Excess Deaths: +33%
COVID-19 Deaths: +50%

In 2020 COVID-19 were 3% lower than excess deaths.
In 2020 +10%!

#COVID19 #CORONAVIRUS #CORONA @justin_hart
Note that both years didn't have any sign. flu deaths, which the baseline contains ('15-'19)!

Calculations here: docs.google.com/spreadsheets/d…
Same chart for the last 12 weeks!

But please note, this is not super reliable yet, as both all-cause excess and covid19 numbers are heavily corrected... (It should be somewhat ballpark reliable though)
Read 4 tweets
11 Oct
Just in the last 12 weeks, all-cause excess deaths have been up by 38% compared to the same time frame in 2020. Whatever we are doing, it's not working!

#Covid #Covid19 #Corona #Coronavirus
Let's look at this by age group:

0-24: -49% (Effect of not having lockdowns?)
25-44: +76%
Read 13 tweets
4 Oct
All-cause excess mortality by week for the USA.

Seniors have a 1st dose vaccination coverage of 94%.
Excess mortality still skyrocketed in that age group.

🤔🤔🤔

nytimes.com/interactive/20…

#Covid #Covid19 #Corona #Coronavirus
Seniors are 94% vaxxed and 83% fully vaxxed.

Also remembering the natural immune!
Here all age groups
Read 4 tweets
28 Sep
Spoiler: I’m working on an analysis that is ground breaking and will flip the whole vaccine narrative around!!! Stay tuned, will post tomorrow. Prob one of my best pieces yet!!!
It’s based on official cdc/study data. Bullet proof IMO
Guys, please stay patient. Still working on write up!

Graphs are finished, but we just made another discovery, that needs to be looked into...

Sorry for the delay...

Still hoping to post today
Read 4 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!

:(