It& #39;s great we& #39;re flattening the curve but what if it doesn& #39;t go back down?

If every happy recovery is replaced by a new infection (R of ~1)...

We could be stuck on a plateau - with a rolling few million people infected, causing 1-2,000 deaths every day for months.

1/ Thread:
First, the optimistic evidence.

Symptom self-report indicates #Lockdown might have reduced # infections by 70% since peak of 1 Apr.

https://covid.joinzoe.com/data ">https://covid.joinzoe.com/data"...

Still implies ~1.5-2m total infected so way too early to ease lockdown - but good trend!

@timspector @Join_ZOE

2/
Next, the charts. Just visually, Italy, Spain, France (and Iran) have seen flattening and a gentle reduction from peak, but then maybe plateaus rather than strong reductions?

Based on this tracker, UK response is still less stringent than Fr/It:
https://www.bsg.ox.ac.uk/research/research-projects/coronavirus-government-response-tracker

3/">https://www.bsg.ox.ac.uk/research/...
UK lockdown weaker than It/Sp/Fr, but they& #39;re still seeing cases rise. Doubled in last 17-22 days ( @OurWorldInData).

Need to remove "recoveries" to estimate live infection levels, but likely flat rather than lowered?

UK doesn& #39;t report recoveries!, so can only model these.

4/
UK case growth has dropped since #Lockdown, but now stuck at ~5-6% per day. Why would it reduce further now we& #39;re a month into lockdown?

I hope doing more testing is hiding a real further slowing.

Also, we are nearing point where daily recoveries should offset new cases.

5/
That still implies we& #39;re holding the real number of infections but haven& #39;t yet begun to reduce it.

Deaths are more reliable (although understated by 1.5-2x given non-hosp / reporting lags / undiagnosed) - but lagged in time.

Also stubbornly stuck at ~6% daily increase.

6/
Most important data is the true % infected.

Random pop sample testing could tell us in days. Govt say they& #39;ve been doing since end Feb but no results yet published.

In meantime, we can estimate from deaths.

@Imperial_JIDEA estimate 4.19% (2.8m) but only use hosp deaths.

7\
My top-down estimate using adjusted death toll suggested 5.5m (8.8%) as of 10 Apr.

Higher concentrations in urban centres.

https://twitter.com/JamieWoodhouse/status/1248661153603084289?s=20

8\">https://twitter.com/JamieWood...
Without understanding true numbers infected, there& #39;s a risk we fail to tighten #Lockdown (may be needed if on plateau) or, even worse, relax lockdown too early.

Doing the former makes our current flat wave longer. Doing the latter makes it even higher.

#HerdImmunity again.

9\
Best way, for people + economy, is:

https://abs.twimg.com/emoji/v2/... draggable="false" alt="1⃣" title="Keycap digit one" aria-label="Emoji: Keycap digit one"> #Lockdown hard enough

https://abs.twimg.com/emoji/v2/... draggable="false" alt="2⃣" title="Keycap digit two" aria-label="Emoji: Keycap digit two"> Relax (carefully) only when we& #39;re sure infected numbers v.low (sample test!) + https://abs.twimg.com/emoji/v2/... draggable="false" alt="3⃣" title="Keycap digit three" aria-label="Emoji: Keycap digit three"> ready

https://abs.twimg.com/emoji/v2/... draggable="false" alt="3⃣" title="Keycap digit three" aria-label="Emoji: Keycap digit three"> Test/trace/quar at scale to spot + stop new outbreaks before we lose control (again)

https://abs.twimg.com/emoji/v2/... draggable="false" alt="4⃣" title="Keycap digit four" aria-label="Emoji: Keycap digit four"> Drugs then vaccine.

10\ END
+ Hospitalisation data also seems more positive re: trend...

https://twitter.com/jburnmurdoch/status/1250904238399598594?s=20

Of">https://twitter.com/jburnmurd... course, says little about non-hospital deaths, but hope it& #39;s still a meaningful indicator of total infections coming down.

h/t @StefanFSchubert + @jburnmurdoch
This pic sums up the risk of an early #Lockdown relaxation beautifully...

Thanks @JKFanghanel https://twitter.com/JKFanghanel/status/1252508742203117568?s=20">https://twitter.com/JKFanghan...
You can follow @JamieWoodhouse.
Tip: mention @twtextapp on a Twitter thread with the keyword “unroll” to get a link to it.

Latest Threads Unrolled: