I want to talk about 'decoupling'.

A thread.

The Government's reason for proceeding with the 17 May Step 3 unlocking was that vaccination means that cases are 'decoupled' from hospitalizations.

Here's the quote in Hansard.

The problem is: it's not clear to me that cases and hospitalizations *have* become decoupled.

It's worth reading this in conjunction with the SPI-M advice available to Government prior to Step 3 unlocking.

SPI-M's projection based on a variant (e.g. Delta/B.1.617.2) being more transmissible.
We know that Delta is around *60%* more transmissible (with a lot of uncertainty).

Anyway, let's look at cases and hospitalizations.

There are three phases.

- before mass testing
- mass testing
- vaccination
I have plotted:
- cases (purple, left hand scale) and
- hospitalizations (red, right hand scale)
Now, these graphs may look similar, but the problem is when there are low case or hospitalization numbers, it's tricky to see what's going on. If hospitalizations were to double from 2 to 4 to 8 we wouldn't see it on this graph, so we need some way of showing doubling.
Let's plot the same chart on an exponential chart. This also allows us to plot the cases and hospitalizations on the same scale.

Hospitalizations certainly appear to have reduced due to the vaccine effect, which is great.
We can see phase 1 - where most of the cases were detected in hospitals, so the hospitalization and case lines are very similar.

Then mass testing where there is a constant-ish distance between the two lines.

And the vaccine effect coupled with a new variant from May 2021.
If cases and hospitalizations were now *completely* decoupled, it would be difficult to find a relationship between the cases and hospitalization lines.

Let's look from the beginning of January to see what is happening.
I think it's pretty clear that hospitalizations are not going up at the rate of cases.

*But* this is to be expected. Becuase there is a delay for cases to be translated into hospitalizations.
Instead of looking at the cases themselves, let's look at the increase compared to the previous week.

Can we see a pattern? But remember, hospitalizations *lag* cases.
So let's shift cases back by 11 days (this takes into account reporting delays and the time over which we have calculated averages).
Now let's just look at the North West of England.
We may be able to get better graphs by plotting by case specimen date rather than reporting date. We could also split by age of cases.

But the main point is that it is too early to say that cases and hospitalizations are decoupled.
For the Government to make decisions on the basis that cases and hospitalizations are decoupled would be reckless.

As ever, these are political decisions.

And we shall see the more detailed modelling on Monday.
*Nobody* wants to go back to another lockdown. The reason this could happen would be (like Lockdown 1, Lockdown 2, and Lockdown 3) that hospitals become overwhelmed and there was a threat to the health system.
The Government promised to use Data not Dates.

And cases and hospitalizations are rising.

• • •

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

Keep Current with Dr Duncan Robertson

Dr Duncan Robertson 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!


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 @Dr_D_Robertson

9 Jun
Here are some charts showing what is going on with Covid cases and hospitalizations in England.

A thread.

Firstly, cases. In order that the graph doesn't get too spiky, we smooth it by taking the average cases in the previous 7 days. Here it is.
We're only really interested in what's happening now (and in the future), so let's look at the right hand bit. Cases are rising exponentially.
To show this, we can plot the case numbers on a logarithmic plot. If there's an exponential rise, this line should be straight.
Read 11 tweets
8 Jun
Let's look back at the SPI-M consensus statement from before Step 3 of the Roadmap.

A short thread.

Read 12 tweets
6 Jun
We are now here.
The source from this chart is Figure 2 of the SPI-M-O consensus statement of 12 May, a few days before we moved to Step 3 of the Roadmap.

The figure comes from the Warwick/JUNIPER Road Map scenarios

Read 11 tweets
3 Jun
Cumulative case numbers for variants published by PHE today, 3 June.

I think we all get the picture now that the Delta/B.1.617.2/'India' variant is becoming dominant.

Here are the charts.

And a final tweet linking to today's risk assessment of Delta, which isn't great news.
Logarithmic cumulative chart.
Heatmaps of weekly cases
Read 7 tweets
3 Jun
The PHE risk assessment of Delta/B.1.617.2/'India' has been updated today, 3 June.

- *new* INFECTION SEVERITY risk assessed as RED (with Iow confidence)
- VACCINES risk still assessed as RED but now with HIGH confidence.

Previous versions linked below.

A thread.
RED risk
HIGH confidence

"Transmissibility appears greater than wild type (first wave) SARS-CoV-2Delta continues to demonstrate a substantially increased growth rate compared to Alpha, across multiple analyses. ...
"... Delta cases are rising whilst Alpha cases are declining. Secondary attack rates, including household secondary attack rates, are higher for Delta, but these are not yet corrected for vaccination status. There is in vitro evidence suggestive of increased replication in ...
Read 19 tweets
3 Jun
What is the 'Nepal Variant'?

This is the Transport Secretary's statement just now, justifying why Portugal has moved from the green list of countries.

"a sort of Nepal mutation of the so-called Indian variant which has been detected"
This was featured prominently in today's Daily Mail.
The World Health Organization apparently do not know about it.

Read 7 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!