Tweet

😀 Stephen Connolly

30 Aug, 11 tweets, 4 min read

Latest Israeli cases analysis, and previous analysis revisited.

First a data health warning. I am basing the numbers in each dose using the published data on ages of vaccinated (data.gov.il/dataset/covid-…).

The readme (data.gov.il/dataset/covid-…) has this caveat. 🤷

(1/n)

With the new data on the 3rd dose I have had to rework the analysis, so the layout will be slightly different than previously. Also to try and limit information overload I only present Z-scores and P-values for selected comparisons against both unvax and 2 dose fully vax

(2/n)

Anyway, here's the data. Interpretation to follow...

(3/n)

1. There is some strange things going on with the case numbers and vaccination numbers around the 3rd dose. There are cases reported as occurring after 3rd dose (e.g. 1-6 days or 7-13 days) for some ages that are reported as having nobody with the 3rd dose!

(4/n)

2. There is a significantly higher case rate immediately following the 3rd dose. This is higher than both the unvaccinated rate and the 2nd dose rate. Two weeks after the 3rd dose and the case rate is significantly lower

(5/n)

3. There is a strange blip in the 0-19 age cohort following the 1st dose. Hard to tease out but might be a similar behaviour as for the 3rd dose infection spike

(6/n)

I think this is probably the most concerning part of the analysis... those who have just got 3rd shot in the past 2 weeks have highest rate of cases. Would we expect them to report at a different rate from the 2 dose only or 3 dose post 14 days groups... I don't think so

(7/n)

Now if the vaccination by age data is not accurate (and the exact meaning of the readme is unclear), those case rates might not be significantly higher than the unvaccinated rate but to do that would erode the reported benefit of the 3rd dose so 🤷‍♂️

(8/n)

Other than that I think I am still forced to conclude that while testing is partially dependent on the proportion of people who get symptoms, the expression of symptoms will largely be the driver of statistically significant differences in case rates

(9/n)

Case rates do not appear so wildly different to support arguments that the vaccine reduces infection (which is not a claim of the vaccine) but the differences we do see support arguments that the vaccine reduces symptoms (which is a claim of the vaccine)

(10/10)

@threadreaderapp

@threadreaderapp unroll

• • •

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

This Thread may be Removed Anytime!

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

More from @connolly_s

😀 Stephen Connolly

@connolly_s

13 Sep

Some interesting questions about the new "active" and "serious" by age graph on the IMoH dashboard. Here's "active" cases and the verticle black line is when they changed the dataset they were reporting

Question 1: What has caused the surge in the population of unvaccinated people that the normalized figures have dropped. If it was that the partially vaccinated were now being counted then we would expect absolute cases for 12-15 to jump by 1400 but they fall slightly... not that

Question 2: Where did the 1400 active cases in 12-15yo disappear to? They're not added to the unvaccinated cases... they're not added to the vaccinated cases... they're not added to the 3rd shot vaccinated cases... these cases have magically disappeared

Read 12 tweets

😀 Stephen Connolly

@connolly_s

10 Sep

Ok. So I finally automated my Z-score analysis of Israeli case data. I have yet to automate publishing the generated graphs to Twitter... but here comes 9 weeks of the same analysis methodology... so you can easily compare. First up 29th August to 4th September inclusive

2021-08-22 to 2021-08-28

2021-08-15 to 2021-08-21

Read 17 tweets

😀 Stephen Connolly

@connolly_s

18 Aug

@RanIsraeli

Here's the update for Aug 8-14th
@RanIsraeli @MLevitt_NP2013 @prof_shahar @OS51388957 @daridor
I do not think my conclusion that its different reporting rates vs vaccination status is contradicted.
At this point I think this dataset on its own is telling all it can tell 1/n

https://twitter.com/connolly_s/status/1425402670232358915

That doesn't mean there are not useful analysis that can be constructed by pairing other data sets with this data set, but certainly anyone saying this data on its own shows any conclusion about rates across vaccination status is IMHO stretching beyond breakpoint 2/n

Also you can use this data to make statements about individual groups, e.g.:
* The case rate in the vaccinated group is not non-trivial, this supports the assertion that "vaccine does not prevent infection"
3/n

Read 7 tweets

😀 Stephen Connolly

@connolly_s

11 Aug

@RanIsraeli

Latest Israeli case data... @RanIsraeli @MLevitt_NP2013 it's complicated, but I'm going to try and explain what it says and what we can conclude. First the background 1/n

So there have been these spreadsheets popping up on social media reporting the age breakdown of cases and the percentage of the population fully vaccinated and highlighting the similarity of those numbers, citing datadashboard.health.gov.il/COVID-19/gener… as the source 2/n

Using that data I was able to compute statistical Z-scores, but as I do not speak Hebrew, my ability to verify the numbers was limited. Those Z-scores implied a significant difference in the case rate between vaccinated and unvaccinated with unvaccinated lower! 3/n

Read 26 tweets

😀 Stephen Connolly

@connolly_s

9 Aug

When facing Z scores of 5 and 7 indicating that the unvaccinated care rate is significantly lower than the vaccinated case rate, the precautionary principle has a lot to say about whether to continue vaccination while we await understanding of the cause of this difference.

For a null hypothesis is that there is no difference in case rates between vaccinated and unvaccinated, the probability of observing a difference like this is 1 in 8110. This is monthly data so that's once in 675 years

The hypothesis everyone is hoping for is that the case rate for vaccinated is less than the case rate for unvaccinated. That requires us to use a single tailed test. The probability of seeing Z=7 on a single tailed test is 1 in 16,207 or once every 1350 years

Read 10 tweets

😀 Stephen Connolly

@connolly_s

8 Aug

@RanIsraeli

Thanks to spotting this retweet on @RanIsraeli 's feed I was able to update my basic Z-score analysis

https://twitter.com/rzioni/status/1423901461679849474

Here's the aggregate for almost the full month. A Z-score of 3.77 is, single tailed P<0.00008, double tailed P<0.00004. The Z-score of 5.28 and 7.00 is where the NCEST tables give up and say it's effectively beyond doubt. These are +ve Z-scores ie saying unvaccinated LESS likely

Caveat: This says there is a difference between the two groups... and it says for the 20-29, 40-49 and 60-69 age groups there is a very significant difference between vaccinated and unvaccinated rates... but it doesn't tell us what the difference is

Read 10 tweets

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!

Share this page!

😀 Stephen Connolly

Try unrolling a thread yourself!

More from @connolly_s

😀 Stephen Connolly

😀 Stephen Connolly

😀 Stephen Connolly

😀 Stephen Connolly

😀 Stephen Connolly

😀 Stephen Connolly

Did Thread Reader help you today?

Like this author's thread?