Israel data: NEJM article
Booster group vs. 2-dose group.

BIG questions on results & methods.

“We considered 12 days as the interval between the administration of a booster dose and its likely effect on the observed number of confirmed infections”
No. You can’t skip 12 days after booster, especially when there is evidence that the risk of infection is increased during the first 1-2 weeks post dose 1.

Reasons for increased early risk: immune suppression, infection at V sites.
Happened in Israel, at both campaigns. Outbreaks in nursing homes.

Actually, they excluded more than half of the infections!

13K infected (Fig 1), only 5,373 analyzed (Table 2).
You can’t get to post 12-days benefit w/o passing through the effect during 1-12 days (possibly harmful).
(Just as you can’t get to 2nd dose w/o passing through 1st dose.)

They report effects from regression models.
But crude (simple, unadjusted) estimates are similar.
Actually, we observe something called “negative confounding”. Adjusted > Unadjusted. Possible, but unusual.
Next, effect on severe cases.
Figure refers to winter wave? (typo?)
About 25% became severe within 1-2 days of PCR+. Strange.
Likely, first PCR+ after admission (for another reason?)
That anomaly observed before. Source of detection bias?
Estimated effects by days since booster.
Do you see a downward trend that was visually saved by day 25 (with a huge CI)?
Is this a true pattern?
Should we not gather more data to see if the downward trend continues and where it ends?
Average over 12-25 days is not good enough
Graphs in supplement also show uncertainty.
Need more data points to conclude.
To sum up:
Considering these questions, does FDA have enough unequivocal evidence to certify 3rd dose?
In the pre-COVID era: No
But this is Newscience.

>>More (bonus 1-3)

• • •

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

Keep Current with פרופ' אייל שחר

פרופ' אייל שחר 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!


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 @prof_shahar

8 Aug
A scary thread on case-reports.
A case-report is part of medical research.
Its weaknesses: very small sample (n=1), no estimated effect.
Nonetheless, it is research.
How come?
It is an observational study and there is “sort of” control.
The observer sees something “unusual”, “surprising”.
Q: what is “surprising”?
A: contrast with expectation
Q: where does “expectation” come from?
A: from past experience.
That past experience is the “control”
Read 6 tweets
8 Aug
They are at least incompetent.
The analysis in the post is valid. What CDC is doing here corresponds to the proportional mortality ratio (PMR) in epidemiology. (You can read about it in the textbook "Modern Epidemiology").
I will quote one sentence from that book:

"An implicit assumption of a proportional mortality study is that the overall death rate for categories other than the ones under study is not related to the exposure".
Translation, for AE= death:

An implicit assumption of the CDC computation is that the overall AE rate for categories other than death is not related to COVID vax. Which means that the effect of COVID vax on all other AE is not different from the effect of other vaccines
Read 5 tweets
7 Aug
Now that estimated VE for severe COVID is down to 70-80% from those false estimates of >95%, we are still left with cardinal questions:

1. Still biased?
2. What is the effect of V on all-cause hospitalization/death?

Still biased?

What are the characteristics of unvaccinated elderly?
Are there shared causes of V status and health outcomes?
Are those unvaccinated elderly (~10%) exchangeable with their vaccinated counterparts (90%) on key health-related variables?
For example, is the proportion of frail elderly similar in those populations?

All these questions convey a single concept in epidemiology: Confounding bias.

Example: if frail elderly make up 20% of unvaccinated but only 10% of vaccinated, true VE is lower.
Read 4 tweets
6 Aug
Cancelled my subscription to Ha’aretz after they rejected my paper on Sweden (published on Medium). (They were scared.)

1. I/others analyzed published VE studies from Israel and pointed to evidence of biases & over-estimation

2. They claimed ~95% VE. False for all endpoints.
3. Little/no VE for infection.

4. And it is not just waning, if true. (Ab levels decline after natural infection, too, but protection is not lost. Memory cells play a role in re-infection.)

5. VE for severe disease is currently 70-80% at most (and still possibly biased).
6. Vaccine has side effects, including deaths. They never report effects on all-cause deaths/hospitalizations. Why?

7. Now 3rd dose vaccination with little to no empirical basis and numerous reasons for concern. Vaccination during a rising wave is a dangerous experiment.
Read 5 tweets
22 Jul
They claimed that lockdown before the peak terminated the rise.

They were wrong. False claim.
Refutation: Sweden & South Dakota (for example).

Now they claim that mass vaccination "decoupled deaths from cases".

They are wrong again. False claim.
Refutation: Sweden & Netherlands (for example).

Let’s see what happened there in recent months.
Here is an impressive case wave in Sweden & Netherlands, which showed up half way through the decline in the winter wave.
Read 7 tweets
13 Jul
Israel: another paper on VE
Count of cases is very confusing:

Trying to split a total of 243 (eTable 1) into mutually exclusive categories of vaccinated and reconcile with a flow chart (Figure 2, with additions).

No clear understanding how the numbers are split.
However, it is clear to me that the data are hiding excess risk between dose 1 and one week after dose 2.

Likely >2-fold, (1.3-fold under the most conservative counting).
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!