Daily numbers, graphs, analysis thread 2

[Barracker]

A follow on thread from here

Please try to keep it data driven, factual and civil.

www.worldometers.info/coronavirus/country/uk/

I'm assuming for the UK the peak in the number of deaths is a better guide than the peak in the number of new infections

This has made me think - when we talk about the UK peaking around Easter weekend, does that mean the peak number of infections or the peak number of deaths?

That's a key question
and we can't tell from official statements

There is a time offset of very roughly 4 weeks between cases and deaths

SAGE would be estimating cases and then a death range over the following weeks
Naturally enough, they would avoid specifying expected numbers & timing publicly, too much chance of reality not following those estimates

Italy are due to peak any day now - although numbers of new infections rose by about 700 yesterday, which is worrying. Hopefully yesterday was just a blip?

I think in Italy when they talk about a peak they mean the number of new infections, so I'm assuming that's what we mean here too.

Can't remember if I've said this on this or another thread - if the number of tests increases each day, the number of cases will also 'automatically' increase, so it may be difficult to tell what the true trend is - the increases will look 'sharper' than they really are. (And the increase in cases won't necessarily be proportionate to the increase in the number of tests, because the populations being tested are going to change, so you can't just pro rata the numbers.)

So does that mean that the trend in cases will involve a lot of estimating (ie estimating the effect of increasing the numbers tested and the change in population)?

Place marking. Finding this thread very helpful. Thanks everyone.

I don’t see how we can judge a peak in infections given we’re in the processes of changing our testing criteria (NHS staff etc.). I thought we were extrapolating estimated infections from the number serious enough to need hospital admissions so adding milder cases skews the figures doesn’t it?

Incidentally does anyone want to give me their take on this graph. I can’t really figure out what it translates to telling us or why there’s a particular peak at beginning of February. (It’s confirmed cases globally excluding china).

cornetto, same point I was trying to make, but you are much clearer/more succint!

I suppose the answer is to continue to measure the trend in cases/infections by the numbers who are hospitalised and test positive, as we have the past two weeks figures for that; and henceforward, separately, by the numbers of nhs workers who test positive - so look at the two groups separately to see the trends in each. And additional groups as they are added. (And then discuss which is likely to be 'best' indicator. obviously more back data for the first group)

Perhaps the only way to truly judge the peak is when numbers of deaths start to decline in that case. Which will probably be early May time if peak number of infections are expected around Easter.

Daily deaths in Italy peaked on 27th March at 919, dropped to 756, went back up to 837 two days ago and were 727 yesterday. Let's hope they stay around the 700 mark today.

Just to clarify - when I say deaths in Italy peaked on 27th March, I mean the peak in terms of numbers we have so far. They could shoot up again to a higher peak in the coming days (though god forbid).

I think we need to distingush peak deaths DAILY cf overall.
When we talk about Italy peaking 3-10 April, they mean total infections since the beginning, peak deaths since the beginning coming later.

When we talk about Italy peaking 3-10 April, they mean total infections since the beginning, peak deaths since the beginning coming later.

Thanks. I really hope we see some encouraging numbers from Italy soon. The elusive peak feels like it's a long time coming, but maybe that's just me.

Peak infections in Italy to come? Wouldn’t they have slowed down since the lock down three weeks ago.

Deaths will start to fall maybe 4 weeks after R0 genuinely drops below 1.0

@BigChocFrenzy what is RO please?
Been following this thread, very interesting and enlightening

R0 is the number of people that on average an infected person will go on to infect. So if it's less than 1, the infection eventually dies out, if it's 1, the number of people infected doesn't increase, and if it's more than 1, the number of people infected increases exponentially, with a greater number infected each day. (The assumes people recover and become infection free.)

R0 can change. For example with social isolation measures or once most of the population have become infected.

R0 (R naught) is the reproduction number.
If it is above 1, each infected person infects more than one person and the disease grows.
If it is below 1, each person infects fewer than one person, and the disease decreases

labblog.uofmhealth.org/rounds/how-scientists-quantify-intensity-of-an-outbreak-like-covid-

Gotcha, thanks 👍

I have a question.
DH pointed out if we test 100,000 people a day (if) it will take nearly 2 years to test everyone.
Assuming NHS staff will need multiple tests, how soon do you honestly think Mr or Mrs Average will get tested?

Spain's sees record death toll, as number of fatalities hits 10,000
Spain’s coronavirus death toll rose to 10,003 on Thursday, up from 9,053 on Wednesday - an increase of 950.

On Tuesday Spain recorded 864 deaths related to coronavirus, the previous highest figure recorded. :( It feels like it will never end here in Spain. :(

Assuming NHS staff will need multiple tests

This. I was thinking about all the political and media angst about testing. But the test only shows if you have Covid on the day you are tested. It is perfectly possible to have the all clear, go in to work, and catch it the next day. Therefore to be effective, it needs to be taken every couple of days! The antibody test is therefore the only way to go surely?

Exactly, but the antibody test will need repeating for anyone testing negative on a fairly regular basis (especially NHS/frontline staff etc) especially as some will have zero symptoms and not know they've got/had it.

Exactly, but the antibody test will need repeating for anyone testing negative on a fairly regular basis (especially NHS/frontline staff etc) especially as some will have zero symptoms and not know they've got/had it

There was a good discussion with a Doctor about how long it takes for antibodies to be detected after an infection. It can take quite a while.

It's also annoying to hear people talk about the antigen test - it's not an antigen test. As far as I am aware, the only test they are using to detect if someone is currently infected is RT-PCR which detects viral RNA. I don't think they have an antigen test to detect the virus ?

The antibody test will tell you nothing useful about medical staff who have symptoms but are not actually infected with Coronavirus. Although I don't know how many people might be in this group.

There's also the question of how long it takes between becoming non-infectious and developing antibodies that can be detected.

Bloody hell pussycatinboots that's a good point!

DH pointed out if we test 100,000 people a day (if) it will take nearly 2 years to test everyone.
Assuming NHS staff will need multiple tests, how soon do you honestly think Mr or Mrs Average will get tested?

Thanks for interesting discussion re: testing. I have nothing of value to add but it’s good to read your opinions and get a better grasp on what testing can and cannot help with.

As PPs point out it will take forever to test everyone and one test per person isn’t enough information to do much with.

With all that in mind I’m inclined to believe that tests need to be reserved for hospital admissions and discharges (CV and non CV) hospital and healthcare workers (and associated key workers such as carers and funeral services staff) AND a properly parametered sampling to see how widespread the virus actually is (not that I know anything about how that would be conducted, but I bet some of you do!)

Tests on demand due to suspicion of mild illness seems like dreamland stuff at present.