Daily numbers, graphs, analysis thread 12

[BigChocFrenzy]

Welcome to thread 12 of the daily updates

Resource links:

Slides & data UK govt pressers
UK dashboard sub-national data, local authorities
Beta Uk dashboard deaths, cases, hospitals, tests, partially sub-national
UK stats updated daily by PHE & DHSC
ONS UK statistics for CV related deaths, released weekly each Tuesday
PHE surveillance report infections & deaths released every Thursday with sep. infographic
NHS England stats including breakdown by Hospital Trust
FT Daily updates
HSJ Healthcare updates
Worldometer UK page
Plot FT graphs compare countries deaths, cases / million pop. / log / linear
Covidly.com filter graphs compare countries
Plot COVID Graphs Our World in Data

We welcome factual, data driven, and civil discussions from all contributors 📈📶👍

Not much change reported in the weekly surveillance report - except the growing proportion of new cases who are Asian - over 40% in the last week, which is at least 4 times higher than it ought to be if everyone was equally at risk. The Leicester outbreak may be partly a cause, but can't fully explain it (at it's not that high a proportion of all cases in Leicester). The proportion of hospital admissions who are Asian is lower than the proportion of cases, suggesting it is young Asians testing positive - maybe to do with employment patterns or a lack of social distancing? (Multi generational households and language barriers being less likely factors if it's young Asians rather than older ones)

Sorry it's a screenshot, but this just popped up on my Sky News breaking:-

647 new cases and 114 new deaths, significantly higher than last Friday

Last Friday was an anomaly. It's the second lowest Friday since 20th March, and looking at day of death it's still running down; 60% of the non hospital deaths happened over a week ago.

I have far more faith in the ONS numbers (which use date of death, and classify as Covid any death where the death certificate confirms the member died of, or is suspected to have died of, Covid). While it was true in March and April that PHE was underreporting deaths, this has not been the case for some time. You can see this when comparing the ONS weekly numbers for Covid deaths with those reported by PHE. For the last available week, PHE figures were 30% higher than the ONS figures. We seem to be the only country now overreporting Covid deaths, and it is unhelpful.

The England NHS analysis shows for the seven days to 11 July the average daily death rate in English hospitals was 21. The weekly ONS reports consistently show hospital Covid deaths at 63.5% of the total. This would put the total deaths per day for England at 33 and allowing for the rest of the UK, probably now around 40.

Today there were 16 new hospital deaths reported in England, but PHE new deaths reported is 114. One does start to wonder where all these are coming from? It must partly be the new issue around anyone who ever tested positive, but it must also be quite high numbers from March, April and May (which are already in the ONS data) being reported now as new deaths. Time for a change in approach in my view.

@PatriciaHolm

Last Friday was an anomaly. It's the second lowest Friday since 20th March, and looking at day of death it's still running down; 60% of the non hospital deaths happened over a week ago.

That's reassuring - I note that the day after last Friday ie 11th July was notably higher 822/ 148

Thank you both, me and my mum we're having a bit of a worry about them, I knew you lot would know the answer though.

more encouraging colour shades on the weekly cases graph from the PHE surveillance report. Compares with two week ago and 28th May which is the weekly rather than cumulative graph I would find in these reports.

Herefordshire is the big red blob and down to the outbreak on the farm, although I see in today's data file a further 14 cases added on the 13th and 27 more on the 15th - presumably down to more test results coming through on the seasonal staff at the farm, as opposed to any outbreaks elsewhere? More confirmed cases in the last six days than they;d had in the previous six weeks. But very localised, and looks to be well captured and contained

big red blob refers to the map published in today's surveillance report - rather more red on the end on May map

which is the first weekly rather than cumulative graph I could find in these reports.

Cooking wine and typing about data - never a good mix :)

@boys3 Yes that would be right about the farm in Herefordshire. It was reported in the news the other day that they'd found a further 19 or so cases, which tested positive later than the initial ones. Given that most of the workers live on site and with each other (so don't infect people working in other places) it should be relatively easy to contain.

Does no-one else pay attention to pictures of people leaving ICU? Patients can easily be in there for over 28 days before they leave, either alive or dead. So if you dont count deaths more than 28 days from a positive test you'll miss some of the sickest people. Still an attempt has been made to quantify that.

As for the idea that after 28 days in the community you have "recovered" from Covid-19 and therefore dont die of it - would someone like to say that to those who got a "mild" case and are still recovering? No-one yet knows how long the effects last or how that varies with age and pre-existing conditions.

Still - case figures now are largely outbreaks in a few location and increases mostly less serious than an increase might suggest. If the saliva tests prove reasonably reliable schoolchildren could be tested regularly making everyone a lot happier about getting schools back. They might even return with no social distancing at all.

I hope someone is paying attention to areas with smaller problems though - we need to stamp on sparks (Hackney and City of London, Braintree - just a couple I noticed) before they become fires.

should have been less serious than higher number suggest - more asymptomatic cases being picked up is actually a good thing.

Very interesting antibody stats from the Isle of Man, thanks to the poster here:

Actual accurate antibody results www.mumsnet.com/Talk/coronavirus/3970661-Actual-accurate-antibody-results#98389340

Six different types of Covid symptoms:

www.kcl.ac.uk/news/six-distinct-types-of-covid-19-identified

Does no-one else pay attention to pictures of people leaving ICU? Patients can easily be in there for over 28 days before they leave, either alive or dead. So if you dont count deaths more than 28 days from a positive test you'll miss some of the sickest people.

Deaths IN hospitals (including ICUs) will be picked up through the daily NHS / PHE reporting. Those numbers do not depend on when the person was tested positive for COVID - yesterday or two months ago. Hospital reporting looks quite reliable and the numbers for England are now down to 20-30 cases a day.

The challenge is with deaths outside of hospitals (primarily in care homes but also in the community). For those cases the 28 day cut off post the last positive test appears consistent with other countries and other UK nations.

@Firefliess

Not much change reported in the weekly surveillance report - except the growing proportion of new cases who are Asian - over 40% in the last week, which is at least 4 times higher than it ought to be if everyone was equally at risk. The Leicester outbreak may be partly a cause, but can't fully explain it (at it's not that high a proportion of all cases in Leicester). The proportion of hospital admissions who are Asian is lower than the proportion of cases, suggesting it is young Asians testing positive - maybe to do with employment patterns or a lack of social distancing? (Multi generational households and language barriers being less likely factors if it's young Asians rather than older ones)

Yes. First thing I noticed was the ethnicity part. Decreasing in white population. Fairly steady around similar figures do black, EM but rising in Asian communities.

I wonder if there's something specific here causing the rise?

I was impressed with Adil Ray on GMB who may a plea to Asian community to follow guidelines due to current figures.

@crazyunicornlady73

647 new cases and 114 new deaths, significantly higher than last Friday

But the gov.uk days no deaths reported due to problems MH has asked to be investigated.

The public health director of Blackburn & Darren, which has also seen a spike, also referred to the ethnic bias in infections there -

"He said there would also be "targeted work" after a rise in infections within the South Asian community - in particular "cluster infections" among families living in small terraced houses".

@itsgettingweird
The problems are basically that someone thought it was a good idea to include the death of anyone who has ever had Covid-19 in the death statistics. This combined with the fact that a lot of the early detected cases were in care homes, where, on the whole, life expectancy is low in normal circumstances (45% of carehome residents die within a year of admission according to one report), means the supposed Covid-19 death toll was beginning to include the deaths of individuals who recovered from the virus months ago and who have now died of unrelated conditions. This methodology would also see the UK continue to add to it's covid death toll for several decades even if the virus was eradicated next week which is clearly nonsensical.

AprilLady I rely on ONS who have excellent data quality and analysis, even when it is for the situation a couple of weeks in the past,
rather the daily official figures and PHE

Yeah I know the issue. It it's that 114 deaths are Ben v reported yet gov.uk have actually said they aren't reporting due to issues.

We've known about people being hit by a bus in June who had a positive test in March being reported as a CV death for ages.

Big choc has reported same for Germany - any positive test is reported as CV death regardless of actual cause.

"The problems are basically that someone thought it was a good idea to include the death of anyone who has ever had Covid-19 in the death statistics"

For PHE

The public health authorities in Scotland & NI followed the rest of the world with the 28 day cut off

Probably the minority of cases still active after 28 days would count as COVID deaths
and PHE did undercount in the earlier days

We should rely on ONS for data & analysis,
rather than taking PHE daily reports too seriously in detail

• and also look at 7-day rolling trends [[https://twitter.com/PHE_uk Public Health England]]d@PHE*_uk

Here's more detail from PHE Incident Director, Dr Susan Hopkins,
on our method for counting #COVID19 deaths.
.......
Of the 40,528 #COVID19 deaths reported by 15th July: < For ENGLAND >

▪️ Around 90% of deaths occurred within 28 days of a positive test

▪️ Of those who died after 28 days, COVID-19 was stated as the main cause of death on the death registration form for 47%

[2/2]

▪️ Counting only those who died within 28 days of a positive #COVID19 test would include 35,664 deaths,
and exclude 4,149 deaths in people with laboratory confirmed infection

< 52% of those 4,149 = 2199 deaths that should be removed from PHE total
So my calc is that 40,528 - 2199 = 38,329 new PHE total >

Then update that total with new deaths

It's fine to say it's only a small number of deaths up until this point, but clearly as more time progresses a larger and larger proportion of deaths will be old, so while I have no reason to doubt that the 10% is a genuine percentage for the entirety I am willing to bet that the figure for deaths between July 1st and July 15th is much higher. What is a good enough methodology in the early days of a pandemic is not necessarily good enough a number of months into it.