Daily numbers, graphs, analysis thread 3

[Barracker]

Welcome to thread 3 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

Thank you to all contributors for their factual, data driven, and civil discussions.

Is it me or are the journos just asking the same questions in 19 different ways day on day at the press conference?

And FFS. Why keep pushing Dominic Raab about whether he can and will make a decision re lockdown continuing on Monday. We have seen all other countries extend theirs by 2-3 weeks and sometimes further.
Are they really expecting him to say "yes, I am now in charge, I will make the decision and it's very likely we'll all be back to normal" 🤦‍♀️

Maybe for total UK deaths we should look at the weekly totals, to try to smooth out reporting blips ?

Well, bigchocfrenzy, if that is the case the other 5100 people are really unwell because at no point during the entire modelling do any of them leave icu. Whereas in fact with reduced road accidents and cancelled operations, plus additional built capacity, there will be more. Of course they will be very busy and no doubt demand will outstrip capacity. But it's farcical to base the whole model on 17k hospital beds and 800 icu beds.

It would be a Cabinet decision, even if Johnson were still in charge

Many people - including their readers - are desperate for lockdown to be lifted
Hence journos are obliged to ask what cannot be answered for some time

This chart uses a 7 day rolling average to smooth out blips, BigChocFrenzy

Yep, Barracker I find the FT chart useful, but difficult to zoom in on - visual disability

A table of the rolling weekly numbers would be a good accompaniment

Callimanco They seem to have used the same system for each country:

assuming that bed occupancy for non-COVID cases stays the same
and did not / could not estimate the new hospitals being built by anyone - probably before their model date

However, it is really the UK that looks the outlier
France, Italy etc look reasonable totals and the US one is lower than other estimates I've read from there

They also haven’t factored in all elective surgery being cancelled and the whole of the private hospitals sector being given to the NHS at cost.

But people were desperate for lock down to happen!

The government have said over and over its a cabinet decision, it'll be based on numbers and they won't know until this week is over what direction we are going. So why ask him the same question 26 times and get the same answer 26 times?
The same as the tests? Matt Hancock was clear yesterday he's not going to meet the target realistically because they still can't get a reliable antibody test. He was asked 3 times today if they are still committed (yes but we've explained why it will probably not be met) and said nothings changed since MH statement yesterday. And then the next person asks.
Do they think they'll all get different answers if they keep asking. It's like what I tell my 15 yo ds. If you don't like the answer don't give me the opportunity to give you an answer you like even less by keep asking

The Chris Giles tweet is a fail

He doesn't understand that he is comparing apples with pears.

The NHS England stats published on 5 April (why is he tweeting this today? update the data) are the NHS death reports to 5pm 4th April and the reports made on 4th April could refer to ANY date up to 4th April.

The gov uk stats are simply 'today's reports less cumulative reports to yesterday'. They can't be anything else because nobody wants to constantly re-evaluate.

He appears to be arguing that '27th March's death reports' is the same as '27th March's deaths' .

27th March's death reports were never 27th March's deaths, they were mostly 25th, 26th, 27th.

NHS England first released the 'date of death stats' on Friday.

AFAICT Wales (Scotland is not that important in that it's not included in 'ONS weekly death all causes' releases) still doesn't do that.

He thinks this is an issue, but it's not. Once you smooth out daily variations due to Sundays etc., the release on Friday is mostly deaths that took place on Tuesday and Wednesday, etc.

It doesn't really matter at all that there is a long tail of old deaths in the reports, as the data from today are 95% within the last 7 days, and 70% within the past 3 days. If you treat it as 'two or three days before publication' then that's more than good enough. It won't cause you to 'miss the peak', you just need to look at trends over three or four days - if they are declining, then we past the peak (deaths) that many days ago plus 2/3 days. Easy.

Much more important in any case is 'excess deaths', which we won't know for several weeks..... His graphs don't show us anything particularly interesting. Just keep comparing daily death stats over a rolling 3 day period. No problem.

ShootsFruitsandLeaves - yes, I can't understand all the ruminating over the minutiae of precisely when the recorded deaths happened. Surely the line of the trajectory over a few days is giving an accurate enough picture. Who is to say any other country has more accurate figures? they must all be subject to the same sort of inconsistencies, especially now that the figures are sadly rather high?

Chris Giles should maybe take a look at my volcano 😉

The actual timestamped data for the 27th is almost complete now - you can see from the volcano chart that there have been 316 actual deaths that occurred on that date.
That number may rise by a small amount, but probably only single figures. It rose by 3 today, by 3 yesterday.

It's looking like the number of new cases per day peaked roughly two weeks after lockdown in Italy and Spain. If our lockdown has been effective then the new cases here should be peaking here about now.

Note that ons registrations in England (1568 dead by 27th march and registered by 1 April) don't contradict NHS stats (1695 by 27th March and reported by 6th April)

Doctor reported deaths are fastest registered - 4 days median, but clearly out of the 300 deaths on 27 you'd expect, say, 100 not to have been registered by 1 April, and a smaller number of deaths for each proceeding day, so there is nothing that suggests a difference between death certificates with covid-19 and nhs stats.

Still ignores the possibility of excess deaths outside the NHS and within it that don't mention covid-19 (nursing homes etc), which will only show up in weeks ahead

NotDavidTennant, we can't measure new cases, only new tests, which are likely NHS staff and ill people mostly. If we scale up new tests we get more positives without any increase in infections.

The IHME modelling is designed to predict how many beds and ICU beds will be needed. It is saying based on current modelling place x needs y more beds and z more ICU beds so free them up or create more.
Does it state anywhere that they are treating lack of availability of beds as a causative factor?

Really depressing news in France this evening. 1417 deaths reported today, and 11059 new infections.

And FFS. Why keep pushing Dominic Raab about whether he can and will make a decision re lockdown

Driving me nuts too, they did the same thing yesterday about shouldn't the PM hand over. Yes but shouldn't he though. But really shouldn't he. I keep waiting for Raab to say shortly, "I've already answered that" and move on. Or hold up a placard. "We Cant Know What The Situation Will Be On The 13th Until We Reach The 13th."

I gather the deaths in France include previous nursing home deaths. It was 602 new hospital deaths. I think not every country is reporting nursing home deaths.

I thought the France figures were jumping around a lot. Today's number is horrific

Yes France has started adding in cases from nursing homes and the figures are rocketing. If you take them out the figures are more stable. Is horrendous though.

I read that care home and nursing home deaths are what have pushed up German death stats the last 10 days or so

Despite all precautions, CV has raged through several nursing homes here
Utter carnage
1000s of ICU beds available, but largely irrelevant when these elderly people are often so frail

The graph posted by Baracker seems to show that we are reaching the peak here. I've got everything crossed that it starts to level out or even drop next week in terms of deaths

The US however still climbing steeply 😔

Of France 10,000 plus deaths more than 3000 of those have only recently been annouced from care homes.