#UKCovid Twitter has been getting excited about anomalies in test results in the SW of England, with LFDs and PCRs producing clashing results. Negative PCRs are leading to previously reported positive LFD cases being removed from the records. It's a mystery! [Thread] 1/15
There has been lots of speculation, which is great if you want to discuss nasal mucus with random strangers on the internet, or want to learn how Bayes' theorem can be applied to theoretical false positives and negatives. But I prefer real numbers. 2/15
A few of us have been studying the weird SW data on the Gov dashboard since 10th September. This thread (the pilot episode of a potential mini-series with even better graphs) serves to show just how odd the situation is. More data will follow. 3/15
For comparison purposes, I want to show the trends in 'normal' data for the past 10 weeks. This graph depicts how positive specimens in London have been falling as a proportion of England's daily cases. During this period, London's share of national cases slowly fell by half. 4/
While London was accounting for a declining share of the English total, regions like the Midlands had significant rises. Driven by surges in Corby, Kettering, and wider Northants, the East Midlands went from about 9% to a peak of 12% of England's daily positive specimens. 5/15
Nine smallish LTLAs in the South West only accounted for 3.5% of English cases on 1st August. Here's a fun game for you. Use your skill and judgement to detect a trend and predict what happened in the second half of this sample. 6/15
Did you see that coming? No, nor did the public health authorities! The reported positive specimens in the parts of the SW near Bristol (what I'm calling the "Suspicious Nine") are really fucking strange. 7/15
Fortunately, the area in question only accounts for about 4% of total cases (much less than full-sized regions), but infections are likely to be rising there and probably quite quickly. 8/15
From this and other data analysis, I'm inclined to think that there were *intermittent* problems with test processing in September, which are now persisting. I can't imagine how faulty LFD batches could cause this ON/OFF phenomenon, but maybe I lack imagination. 9/15
My working hypothesis for several weeks has been that the issues are on the PCR processing side of the system. I could be quite wrong, but I will post more data in support of this theory in due course. 10/15
Meanwhile, UKHSA is investigating. I just wish they'd realised a problem existed over a month ago, like some of us on Twitter did, as I fear the real-world consequences of giving incorrect test results for 35+ days are very serious. Transmission could be surging "invisibly". 11/
Here's a Guardian article about the issue, with some hypotheses, informed speculation and some advice for those that have received clashing test results: theguardian.com/world/2021/oct… 12/15
I've discussed this issue casually with several respected twitter people, but I want to tip my hat to @OliasDave and @UncleJo46902375 in particular for their help in hunting down data and building spreadsheets/graphs that helped me to understand what's happening. 13/15
This was the eye-opening visualisation posted by Dave a week ago that greatly inspired this particular thread: 14/15
I might not have much time to answer questions, and I'm not particularly interested in speculating on the *precise* causes of the data anomalies. I just wanted to illustrate the strangeness. Explanations may come later, but probably not from me! [END OF THREAD, for now.] 15/15

• • •

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

Keep Current with Someone on the internet 3.8 🥒

Someone on the internet 3.8 🥒 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 @ArtySmokesPS

11 Oct
Just over a week ago, I had a bad feeling about Ipswich (don't we all? - ed) when it was ranked 7th in the East of England for weekly Covid rates. It's really kicking off there now, with daily specimens close to its January 2020 high-point.
*Tips hat to NoDPI sigma.*
Graphs via @Ian_Campbell_ as usual. East of England last 6 weeks weekly total rates: iancampbell.co.uk/covid-19.php?s…
January *2021* I meant. I'm never quite sure if we're in year 1 or year 38 of the pandemic. :/
Read 4 tweets
21 Sep
I don't know what they're sprinkling on the kids' Weetabix, but Kettering has overtaken local rival Corby to go #1 in the UK Covid charts. Local schools are full of it and positive specimens just hit an all-time high. #EastMidlandsForTheLose ImageImageImageImage
Image
Similar things happening in parts of West Midlands that haven't really hit the headlines before. Quite a separation in case rates between under- and over-60s developing, as Covid spreads through school-age humans. #Rugby #Stafford ImageImageImageImage
Read 14 tweets
5 Sep
**Covid FESTIVALS final wrap-up THREAD**
Thousands of tents were abandoned at Bank Holiday festivals, but did campers take Covid home with them? Let's have a look at the graphs and maps. TLDR; Covid transmission at big BH festivals was not as bad as some had feared. Image
CREAMFIELDS (Halton, Cheshire).
Daily rates (by day of report) went up in a few LTLAs near to the site in the days following the festival, but the weekly rates remained below those of other parts of the NW. ImageImageImageImage
Creamfields was 18+, but cases rose faster in Blackpool, Allerdale, Tameside, and Barrow, where outbreaks in unvaxed 15-19yr olds were probably driving numbers upwards. (This may be investigated in another thread). ImageImage
Read 17 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!

:(