Sebastian Funk Profile picture
Sep 16, 2020 8 tweets 3 min read Read on X
Much attention has recently focused on increases in the reproduction number R across the UK, with some speculation on the impact that reopening schools may have had. But for that it is too early to tell. 1/8
First, the numbers. According to latest SAGE estimates, R in the UK and England was between 1 and 1.2. gov.uk/guidance/the-r… According to the REACT infection survey, the reproduciton number in England was between 1.4 and 2.0. imperial.ac.uk/medicine/resea… 2/8
Our latest @cmmid_lsthm estimate of the reproduction number in England, based on today's data, was between 1.1 and 1.4. epiforecasts.io/covid/posts/na…
All these estimates use slightly different methods and data but all point at the same trend: increasing numbers of infections. 3/8
So when did R cross 1? In most regions it probably did so a while ago. Take London, for example. Our estimates from test-positive cases point at an R of 1.1-1.2 and have done so for at least two months. These numbers are consistent with a steady exponential increase in cases. 4/8
Importantly for interpreting these estimates, today's case data only tell us about new infections (and therefore R) until the end of August. That is because infections take about 5 days to cause symptoms, plus 4 days until a test result (if done right after symptom onset) 5/8
These test results then take another few days to enter the daily data. In other words, any observed increases in R do not yet reflect schools reopening in early September, let alone the changes brought in this week. 6/8
This also means that we would expect cases to grow for at least another two weeks just catching up with infections that have already happened. However, if fewer people can get a test, this could cause a decrease in daily case numbers even if infections have been increasing. 7/8
In summary, R was almost certainly greater than 1 in most of England before schools reopened. It will take another 2-3 weeks before we see any potential effect of this Monday's policy changes. In the meantime, it's on all of us to minimise potentially infectious contact. 8/8

• • •

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

Keep Current with Sebastian Funk

Sebastian Funk 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 @sbfnk

Apr 27, 2022
We’ve been investigating Ct values as a proxy for viral load of symptomatic (pillar 2) Covid-19 cases in England in the last 18 months. Work-in progress report (w/ @seabbs) at epiforecasts.io/covid19_ct_pil… and a few observations in this thread. 1/9
First, a caveat. This is all based on pillar 2 testing and therefore affected by test-seeking behaviour. We know about reported dates of symptom onset but not how severe symptoms were, and/or what prompted people to seek a test. This can bias the results shown here. 2/9
Second caveat is that Ct values don't correspond one-to-one to viral loads (and can be affected e.g. by sample site and quality), let alone infectious viral loads. 3/9
Read 10 tweets
Mar 7, 2022
Many now focus on the amazing prevalence data published weekly by ONS for tracking SARS-CoV-2 in the UK. There's a few additional epidemiological quantities that can be derived from it, which we're putting in a weekly report that is based on the ONS data. w/ @seabbs 1/9
First, incidence of infections. Using previous estimates by @HellewellJoel et al. we relate positivity to new infections, taking into account that it can take a few days from infection to PCR positive, and probability of testing positive varies over the course of infection. 2/9
We use the estimates of infection incidence to calculate their growth rates and, by combining them with estimates of the generation interval distribution, reproduction numbers. 3/9
Read 9 tweets
Jan 4, 2022
Lots of speculation on generation intervals of omicron and whether it might different to ones of previous variants. So here are a few things that I think are worth bearing in mind. 🧵 1/16
We can think of the generation interval, i.e. the time between becoming infected and infecting others, as being made up of two components. One is given by how an individual’s infectiousness changes from the time since infection. 2/16
The shape of this curve can depend on age, immune status, variant etc, and probably so do generation intervals. Fig. from study by @terrycojones @GuidoBiele @c_drosten et al. github.com/VirologyCharit… 3/16
Read 16 tweets
Mar 4, 2021
Evidence from the UK testing programme in educational settings suggests that rapid test specificity exceeds 99.9%. A brief thread on how we found this and why 99.9% isn’t necessarily enough. Short note (w/ @StfnFlsch) available at cmmid.github.io/topics/covid19… 1/7
Every time a large-scale survey is conducted with a diagnostic test one can calculate a lower bound of test specificity by assuming all positive test results were false positives. 2/7
Using this on the reported data points for LFD test use in educational settings in the UK we find that with 95% confidence rapid test specificity exceeds 99.93%. 3/7
Read 7 tweets
Jan 8, 2021
Is the new variant of SARS-CoV-2 more transmissible than others and if yes by how much? We @cmmid_lshtm are currently looking at this by comparing reproduction numbers between local areas in England (w/ @seabbs). WORK IN PROGRESS, NOT PEER REVIEWED. 1/10
raw.githubusercontent.com/epiforecasts/c…
Looking at this week-by-week, there seems to be a trend where areas that see more S-gene target failure (SGTF, a proxy for detection of the new variant) have higher reproduction numbers. WORK IN PROGRESS, NOT PEER REVIEWED. 2/10
However, this is confounded by differences in policy (tiers) as well as, potentially, behaviour, demographics, etc. We are using a range of statistical models to try and correct for these factors using data on local policies and mobility. WORK IN PROGRESS, NOT PEER REVIEWED. 3/10
Read 10 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

Don't want to be a Premium member but still want to support us?

Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal

Or Donate anonymously using crypto!

Ethereum

0xfe58350B80634f60Fa6Dc149a72b4DFbc17D341E copy

Bitcoin

3ATGMxNzCUFzxpMCHL5sWSt4DVtS8UqXpi copy

Thank you for your support!

Follow Us!

:(