Thread by @USMortality

Thread on how to calculate Vaccine Efficiacy (VE) as defined by CDC, and deep dive into MA Barnstable County MWWR outbreak case data.

⬇️⬇️⬇️ 1/n

#Covid #Covid19 #Covid_19 #Corona #Coronavirus

In order to Calculate VE, we need to calculate Risk Ratio (RR).
CDC, in their epidemiological handbook, gives this definition/example for Varicella:

Then we can calculate the VE as follows:

I've rebuilt above example in Google Sheets:

- VE for this varicella example/outbreak is at 73%.
- Not statistically significant at 90%, but close.

Now we should be able to do the same for Covid19 right?

I'm using the CDC MWWR Case Study of the Massachussets/Barnstable County outbreak, as that had quite a lot of data.

Unfortunately they do not disclose the # of negative tests per group - so we can try to estimate them.

So we know that about 60k people were in the county.
CDC also discloses the % of vaccinated/unvaccinated.
So we can calculate the absolute size of the two groups.
Now, the question is, what's the proportion of people in each group that got tested? We don't know.. let's use 10%

Using 10%, gives us the total # of tests, and we can finally calculate VE.

VE is calculated the same way as before. So what's the result? --> -26%

So if the proportion of people tested is equal, the VE is negative! WOW

So to get to a CDC defined minimum VE of 50%, the amount of people tested in the vaccinated group would have had to be at least 2.5x times (10% vs 4%) higher.

Is that realistic?

So unfortunately CDC does not know or publish the # of tests by vaccine group.

What do you think?
Please let me know your thoughts in the comments?

Spreadsheet/Sources here: docs.google.com/spreadsheets/d…

FYI, I'm planning to FOIA the numbers from MA or CDC.. So I'd appreciate to get some thoughts before ;)