~2.4 million 🇬🇧 UK people may have #COVID19.
~3.6% or 1 in 28. ~8% in London.
Assume: 1228 dead; IFR 0.9%; Adj. days to death ~18.
So ~136k infected 18 days ago
~4.25 days to double so ~4 doublings
So infections to date: ~2.6m, ~5% recovered so ~2.4m infected.
~3.6% or 1 in 28. ~8% in London.
Assume: 1228 dead; IFR 0.9%; Adj. days to death ~18.
So ~136k infected 18 days ago
~4.25 days to double so ~4 doublings
So infections to date: ~2.6m, ~5% recovered so ~2.4m infected.
Main changes:
Yesterday I reduced median days to death from 18.5 to 13 to account for exponential skew. Now updated to use days to death from infection (23), not from symptoms. Brings adj. number back to 18.
Used slower doubling assuming lockdown is shifting towards Italy (7d).
Yesterday I reduced median days to death from 18.5 to 13 to account for exponential skew. Now updated to use days to death from infection (23), not from symptoms. Brings adj. number back to 18.
Used slower doubling assuming lockdown is shifting towards Italy (7d).
Whole model is v.sensitive to inputs, so low confidence + wide ranges apply.
More deaths --> more infections
Higher IFR% --> fewer infs (don't need as many for same deaths)
Longer days to death --> more infs (more time to double)
Higher doubling rate --> more infs
More deaths --> more infections
Higher IFR% --> fewer infs (don't need as many for same deaths)
Longer days to death --> more infs (more time to double)
Higher doubling rate --> more infs
Model is also v.top down. Doesn't reflect regional / demographic variations.
London % is based purely on 31% share of UK.
For a more detailed model that uses differential IFR% for London + regional breakdowns...
telegraph.co.uk/news/2020/03/2…
From @edge_health_
London % is based purely on 31% share of UK.
For a more detailed model that uses differential IFR% for London + regional breakdowns...
telegraph.co.uk/news/2020/03/2…
From @edge_health_
London estimates are based on yesterday's 31% of UK "confirmed cases" - awaiting breakdown for today.
Implies ~662k current infections in London, or ~7% of the population.
Within the range mentioned by @neil_ferguson in S+T cttee:
Implies ~662k current infections in London, or ~7% of the population.
Within the range mentioned by @neil_ferguson in S+T cttee:
I'm using "Adjusted days from infection to death".
Median days infection to death is ~23.
However, given exponential growth of infections, a higher proportion of deaths to date were caused more recently than that. Hence adjusted to ~18 days.
Thanks to @cheianov!
Median days infection to death is ~23.
However, given exponential growth of infections, a higher proportion of deaths to date were caused more recently than that. Hence adjusted to ~18 days.
Thanks to @cheianov!
Gov data on "recovered" cases has been stuck at 135 for days now. I est 2.3%.
arcgis.com/apps/opsdashbo…
Given "cases" are only the most serious (we ~only test in hospital) I've used higher recovery rate 5%.
Given exponentials, vast majority of those ever infected still are now
arcgis.com/apps/opsdashbo…
Given "cases" are only the most serious (we ~only test in hospital) I've used higher recovery rate 5%.
Given exponentials, vast majority of those ever infected still are now
The better the NHS, the higher the "days from infection to death".
That gives infections more doubling time since the infection point, meaning there are many more current infections for each actual death to date.
If disease kills quicker, there are fewer infections now.
That gives infections more doubling time since the infection point, meaning there are many more current infections for each actual death to date.
If disease kills quicker, there are fewer infections now.
Also maybe counter-intuitively, the more deadly the disease is (infection fatality rate %), the fewer infections were needed 2-3 weeks ago to have generated the deaths you see today.
If it's less deadly, there must have been many more infections to drive the deaths.
If it's less deadly, there must have been many more infections to drive the deaths.
If we were doing population sample testing (as promised here) gov.uk/government/new… we could observe the infection doubling rate.
As nothing has been published, I've used lagged death + case doubling rates + assumption we're starting to follow Italy's flattening.
As nothing has been published, I've used lagged death + case doubling rates + assumption we're starting to follow Italy's flattening.
This would imply total UK death toll of ~23k...
But only if IFR remains 0.9% + we see zero new infections.
Sadly, neither of the above are sound assumptions.
If our ICU capacity is overrun, the IFR% will spike hard.
New infections depends on hard #Lockdown + test/trace.
But only if IFR remains 0.9% + we see zero new infections.
Sadly, neither of the above are sound assumptions.
If our ICU capacity is overrun, the IFR% will spike hard.
New infections depends on hard #Lockdown + test/trace.
A wave is crashing over our ICUs.
We can't reduce the size of that wave because it's caused by infections from 2-5 weeks ago.
We can improve our readiness by scaling ICUs + protecting NHS staff.
We can stop the wave getting bigger + longer by hard #Lockdown + test/trace.
We can't reduce the size of that wave because it's caused by infections from 2-5 weeks ago.
We can improve our readiness by scaling ICUs + protecting NHS staff.
We can stop the wave getting bigger + longer by hard #Lockdown + test/trace.
A more optimistic view? If #Lockdown has already helped slow infection doubling over last 3 weeks to every 5 days, takes # infections down to ~1.6m.
Still projects 15k death toll assuming ICUs not over-run and zero new infections from today.
Neither are sensible assumptions.
Still projects 15k death toll assuming ICUs not over-run and zero new infections from today.
Neither are sensible assumptions.
It's tempting to think... "There's 1228 deaths and #Covid19 kills about 1 in 100, so there must be ~136k infections".
The problem is you're nearly right. But those ~136k infections were 2-3 weeks ago. That's how long it takes to kill.
They've been doubling every 3-4 days since.
The problem is you're nearly right. But those ~136k infections were 2-3 weeks ago. That's how long it takes to kill.
They've been doubling every 3-4 days since.
This isn't the best way of estimating % infected!
We#re waiting for people to die, assuming fatality rate to judge how many infections drove those deaths, then using doubling rates to judge infections today.
A better way? Population sample testing!
We#re waiting for people to die, assuming fatality rate to judge how many infections drove those deaths, then using doubling rates to judge infections today.
A better way? Population sample testing!
Why is estimating # infected more important than just tested "cases"?
1⃣ Infected people are doing more infecting. The "cases" are mostly in hospital or dead.
2⃣ If you know # infected and # deaths you can assess how deadly this thing really is (IFR%, not just "case" fatality)
1⃣ Infected people are doing more infecting. The "cases" are mostly in hospital or dead.
2⃣ If you know # infected and # deaths you can assess how deadly this thing really is (IFR%, not just "case" fatality)
3⃣ Telling public % infected helps us comply with hard #Lockdown.
BJ's speech should have started:
"There are infected people in every train, in every shop, in every park, in every tube... you might even be one of them... please take what I am about to say very seriously..."
BJ's speech should have started:
"There are infected people in every train, in every shop, in every park, in every tube... you might even be one of them... please take what I am about to say very seriously..."
4⃣ "Total infected" tells us how many people are going to be coming through our ICU capacity in the next few weeks and whether that will be overrun.
If it is, #Covid19 will kill much more than 0.9%.
If it is, #Covid19 will kill much more than 0.9%.
Instead, government, advisors and press continue to talk about the ~19k "confirmed cases".
I've only seen 2 newspaper articles (1 ES 2 days ago, 1 Telegraph today) that even mention total infected estimates.
These are just the tip of a very large, deadly iceberg.
I've only seen 2 newspaper articles (1 ES 2 days ago, 1 Telegraph today) that even mention total infected estimates.
These are just the tip of a very large, deadly iceberg.
If we'd started estimating % infected curves from day one (ideally via. regular pop. sample tests), we might have acted sooner.
We might also have persuaded people to comply better with our weak #Lockdown.
Instead... "a national scandal" @richardhorton1
thelancet.com/journals/lance…
We might also have persuaded people to comply better with our weak #Lockdown.
Instead... "a national scandal" @richardhorton1
thelancet.com/journals/lance…
A summary of the model here if it's more readable: medium.com/@jamie.woodhou…
We can easily re-run this for other countries. Let me know if interested.
We can easily re-run this for other countries. Let me know if interested.
