Can we trust the numbers from the UKHSA?

Ugh. Despite living on the other side of the Atlantic, much of my analysis has been focused on the data in the UK because the availability and quality was much better than the data in the US. Or at least is appeared to be. As I’ve been doing some additional analysis of the Surveillance reports and the pregnancy data in particular, I am seeing some things that make me concerned about the reliability of the published data. Since I have many data-savvy readers, please check me on my calculation and logic.

Sums that don’t match

In the 2022 Week 8 report, we see the following:

• 4,780 women 1st dose prior to pregnancy

• 8,353 vaccinated in 1st trimester

• 28,468 vaccinated in 2nd trimester

• 39,138 in 3rd trimester

• 22,887 known vaccinated, unknown trimester

The total of those numbers, which should cover all women receiving at least one dose of vaccine prior to giving birth is 103,626 which is significantly higher than the value of 78,759 which is the number given in the 1st sentence of the paragraph. The same problem presents in the report when data is next updated, Week 12.

Implausible implied data

The 1st update of the pregnancy data occured between the 2022 Week 3 report to the 2022 Week 4 report.

Week 3 tells us:

Week 4 updates:

This says that Jan-Aug had a stillbirth rate of 3.60 per 1,000 & Jan-Oct had a rate of 3.90 per 1,000 for vaccinated women. This jumped out to me as a big increase in the aggregate rate based simply on adding two months to the data set. Sept & Oct must have been really bad in order for the year to data to increase that much in such a short period of time. How bad? Well, we can calculate by using the monthly numbers on births & vaccination status, plus some logic and arithmetic. The other piece of data we need is the following:

So, now we can calculate the following:

• Jan-Aug

  • 324,749 unvaccinated women gave birth

  • Using a rate of 3.60 stillbirths/1,000 → 1,169 stillbirths

• Jan-Oct

  • 382,591 unvaccinated women gave birth

  • Using a rate of 3.90 stillbirths/1,000 → 1,492 stillbirths

• Sep-Oct

  • 57,842 unvaccinated women gave birth

  • Stillbirths: 1,492 (Jan-Oct) - 1,169 (Jan-Aug)= 323 stillbirths

  • → 1000 x (323/57,842) = 5.58 stillbirths per 1K

This would be a deeply concerning number if true. We are left with two possibilities.

1. A mistake was made in the reports and the numbers are wrong (I suspect this is the case, and the stillbirth rate Jan-Aug is reported too low)

2. Something really scary happened to unvaccinated women in Sept-Oct in their pregnancies. If the data is accurate, this should be investigated immediately.

Again, peer review is encouraged.

Update: UKHSA responds to 1st question

UKHSA answers one of my questions