Daily numbers, graphs, analysis thread 10

[BigChocFrenzy]

Welcome to thread 10 of the daily updates.

Resource links:

Worldometer UK page
Financial Times Daily updates and graphs
HSJ Coronavirus updates
Johns Hopkins Coronavirus Resource Centre
NHS England stats, including breakdown by Hospital Trust
Covidly.com to filter graphs using selected data filters
ONS statistics for CV related deaths outside hospitals, released weekly each Tuesday

We welcome factual, data driven, and civil discussions from all contributors 💐

I appreciated some graphs on Sky News this morning.

Recessions compared and sectors.

The drop in the recessions one is stark but I do think Sunak knew this and the busting financial support matches it.

It’s what happens now and next as this support ends which is crucial.

11 is a really tiny number.

From what I can see the upper bound reflects not only the small sample size but also geography and clustering (they tested multiple people from the same house) - because it would otherwise be more like 0.03-0.1%, not the 0.02-0.12% stated, for a 95% confidence interval.

Increasing sample sizes as a blanket policy won't be that useful, because in general sampling error decreases with the square root of the sample size, so you'd need to quadruple testing to get to 0.06%+-0.02%, rather than 0.06%+-0.04%.

The fact that 19,922 people were NOT positive is itself incredibly significant information.

Likely more useful than just increasing testing would be to test-bomb around the infected individuals - WHERE were the 11 people positive.

Because you can have 11 people positive in the whole country, but then it turns out all 11 were in the same town, so the infection rate could just be 10% in one area, and 0.00% everywhere else..... (as an extreme example)

I don't know if I've missed this upthread but the Zoe study also shows falling cases. It indicated a 47% drop in daily cases in the community.

covid.joinzoe.com/post/covid-cases-uk

The results are based on over a million people using the app. 12,872 carried out swab tests after an invitation from ZOE when they began to feel unwell and 108 tested positive. I don't know if there's any testing for asymptomatic people, it doesn't look like it reading the report, however I was called for a test with a mild hay fever-related sore throat. .

"People won't know the passengers they sat / stood near on their public transport commute"

Contact trackers in China and some other countries have tracked contacts on public transport by asking infected people about their commute details and then using season tickets, station cameras etc

This is where an App should be much faster though

An heavily surveilled dictatorship is always going to have an advantage when it comes to contact tracing and tracking

Hi
Recently name changed but have been an occasional poster on these threads under my previous name.
Just seen that the ONS has updated the data on the postcode-level death rate map, to the end of May. My area has sadly increased from 2 to 6 deaths.

www.ons.gov.uk/peoplepopulationandcommunity/healthandsocialcare/causesofdeath/articles/deathsinvolvingcovid19interactivemap/2020-06-12

wintertravel Your analysis of Bergamo seems to take some rather optimistic estimates:

Looking at the UK excess deaths 60k vs COVID 40k, the ONS has stated that some of the gap will be undiagnosed COVID, especially among the elderly and some will be conditions exacerbated by COVID.
Deaths from RTAs etc have reduced

The NHS became the National COVID Service, so I would expect a similar % of deaths due to untreated non-COVID ailments as in Bergamo.
Also, I'm not sure whether there could be a significant spike in cancer deaths during those 10 weeks or so - it would be spread over the next couple of years.

"Bergamo COVID deaths - the 6,000 number quoted in the Euronews article appears to represent excess deaths rather than COVID related cases.
The official COVID number seems to be closer to 3,000-3,500"

Reports from mayors in N Italy are that the number of COVID and other deaths have still been underestimated,
so the number of excess deaths may increase further, due to the disruption & delay there
Especially elderly people living along may still to be discovered - doctors have reported finding such deceased in the UK

So while 6,000 may be an over-estimate, it is likely much more accurate than the 3,000 - 3,500 official figure

1. Age difference however, is significant to extrapolate from Bergamo to UK,

although I can only find the median age Italy vs UK atm and I would like need to compare the % of say age 65+ in specifically Bergamo vs UK

"Of course, there are scientists who still disagree with this but Bergamo numbers do not necessarily support their case."

The Bergamo numbers are evidence supporting the higher number, just as other analyses can be evidence supporting a lower number.

"CDC seems to quote 0.4%"

The CDC low estimate is highly disputed

CDC would mostly have figures from the US, which would include NYC - which has higher IFR & CFR - but also US states with very low population density, totally unlike the UK

Also ....
Public health experts are accusing the CDC of bending under political pressure to say the coronavirus is less deadlyy_.

New CDC estimates of coronavirus death rates look suspiciously low and present almost no data to back them up,
say public health experts who are concerned that the agency is buckling under political pressure to restart the economy.

UPDATE (5/28):
An epidemiologist at the Johns Hopkins Center for Health Security also contradict the CDC's low numberss_, saying that after reviewing new widespread testing,
"The current best estimates for the infection fatality risk are between 0.5% and 1%," according to NPR.

I've been in favour of restarting schools and the economy weeks ago,
but it should not be done by claiming the lowest estimates of risk are the only ones worth considering,
or by claiming the dead would have died anyway in a few months

Instead, we should make the case that children are at almost no risk of serious harm,
but are at serious risk of having their education damaged and their future career prospects affected

and that whatever the fatality level for adults
the economy must be restarted soon, if we are to retain a functioning welfare state and public services

Lockdown was necessary for the intiial period when we were unprepared and knew very little about COVID,
but much less drastic measures should suffice now that infection levels are so low

11 people infected out of 19,000 is TINY
and govt decisions about what happens next should reflect this

Hi @EinsteinsFineWine

The underlying data are here

www.ons.gov.uk/peoplepopulationandcommunity/birthsdeathsandmarriages/deaths/bulletins/deathsinvolvingcovid19bylocalareasanddeprivation/deathsoccurringbetween1marchand31may2020

It isn't possible to map on a postcode level. Rather your postcode maps to a MSOA, which is a geographical are of 5000+ people.

I don't think I approve of their visualization (attached) as if for example an area contains 100 people and 1 million sheep and all the people died, that won't show in the map, since said area will be very large. I think makes more sense to either geographically equalise the MSOAs but still have them vaguely resemble the UK (if that is possible!?) Or just colour code the death counts and then paint the entire MSOA the relevant colour.

"the mortality rate fell by more than a half between April and May in two-thirds of local authorities and the greatest decrease was in Tower Hamlets where the rate fell by 92.5%."

I wonder if they will re-do the ethnic minority statistics now that the profile of those being infected has fundamentally changed from the early urban Londoners to old white people all over the country.

BigChocFrenzy

I have used ranges for Bergamo death numbers since, as you have said, while 3,000-3,500 is definitely low, 6,000 might be high. I was questioning the 6,800 number estimated earlier.

If we use - say - 5,000 deaths and assume that 57% of 1.1mm Bergamo population have been infected, this gives an estimated IFR of 0.8%. The number for the UK should arguably be lower because the UK population is younger. Ferguson's 500,000 estimate calculated based on 0.9% IFR does look reasonably conservative in this context.

@wintertravel1980

Most of the recent papers I have seen estimate COVID IFR to be closer to 0.5%. CDC seems to quote 0.4%

I hadn’t sufficiently factored in age... even 5 years makes a huge impact. As for excess deaths versus confirmed Covid deaths.... I’d strongly suspect that a significant number of the excess deaths that weren’t identified as Covid, were actually Covid as has been mooted recently in Italy, though I accept not the full 6,000.

The IFR can never be known precisely be different is different countries - and is largely theoretical given that suppression and mitigation are in place. The lower CDC figure of 0.4% is one calculated in settings where significant mitigation/suppression activity has occurred in most part, so likely lower than in a “business as usual” scenario that can be used to compare on a like-to-like basis with common respiratory illnesses such as flu.

The most important part of the Bergamo study to me was 57% with antibodies... That’s a huge proportion and has all sorts of implications for understanding of Covid.... not notably that a large majority people (as far fewer than 100% will have been infected even in Bergamo!) do have an antibody response if exposed in an uncontrolled environment, and that T cells aren’t sufficient on to confer immunity for most people, but only a degree of protection.

not notably that a large majority people

should have been

notably that a large majority people ... not sure how the random “not” got in there.

T cells aren’t sufficient on to confer immunity for most people, but only a degree of protection.

Yes, I fully agree with this. My personal hypothesis has always been that innate immunity and T-cell protection may work in cases with mild to medium viral load. When infection is out of control, people might be less likely to just "shake C19 off" and end up with more severe cases of illness.

We need to be cautious about that Bergamo study. There is some conversation on Twitter about it right now, amongst people who have actually found the original study in Italian, and it would appear that it is not a random study - they sampled both healthcare workers and civilians, and in the civilian sample, almost all were already in quarantine; so they had already knowingly either got, or been exposed to, someone with Covid. It's not a random sample of the population.

Which would suggest the prevalence of antibodies in the entire population would be somewhat lower - the Mayor of Bergamo told journalists yesterday a more likely figure would be 45% .

"The most important part of the Bergamo study to me was 57% with antibodies"

I agree
It's surprising
without lockdown presumably even more would have been infected

So herd immunity - in Italy at least - needs much higher than the 60% suggested.

However, that 60% always seemed more based on hope than evidence

Re: The ONS postcode map

Does this record the home address of those who died? Or their place of death?

Im assuming it must be the home address but the webpage doesn't explicitly state this.

@PatriciaHolm

Thanks. I hadn’t appreciated that... 45% is still very high.

What is also interesting is the IFR - for the same age group - in the different areas:

Also - I'm assuming geographical wards with a large number of carehomes will show as having more deaths.

My personal hypothesis has always been that innate immunity and T-cell protection may work in cases with mild to medium viral load.

It’s born out of everyday experience too... When I’m run down, over-tired or stressed, I’m more likely to come down with something.... implying that natural immunity isn’t binary. I know antibodies don’t provide 100% protection either, but it seems a lot more clear cut.

Put 100 people in close proximity with someone suffering with mild Covid symptoms for a few hours and a proportion may get ill. Put the same 100 people into a Covid intensive care ward with no PPE for a few hours, and I expect a lot more would become ill!

"Ferguson's 500,000 estimate calculated based on 0.9% IFR does look reasonably conservative in this context."

WIntertravel Whitty and Vallance referred to a similar number as the "reasonable worst case" scenario

Obviously such a figure has to be conservative, or it becomes "most likely scenario" or even "most optimistic scenario"

Any responsible government has to take action when the CMO states a reasonable worst case

However, now we are far past that, so policy needs to reflect this
but instead seems to have become frozen for weeks longer than necessary

• maybe because lockdown time has not been used efficiently and systems like contact tracking have taken so much longer to organise than elsewhere.

@BigChocFrenzy

Interesting. It seems the 50-59 death rate is similar to the all age flu death rate for a very bad flu season (ie 0.1% - the approx rate for 2017-18 flu season)

Ferguson's 500,000 estimate calculated based on 0.9% IFR does look reasonably conservative in this context.

The important thing for public health policy here is the order of magnitude, not whether it was 20-30% out.... If the figure was 300,000 or 700,000, I don’t suppose it would have, or should have, made a difference in terms of public policy response.