Number of Deaths

[Harry Wilson's tackle]

What do you reckon will be the effect of the new UK numbers on your trends ? I have noticed that all the historic figures have now been adjusted upward to reflect these new 'all settings' figures. I've got all the original figures recorded, but it seems as if all graphs etc have now been reworked on these 'all settings' numbers, so there is no stepped increase from 5 days ago. What I don't understand is that if we had these figures all along, why didn't we use them

However, I believe the new 'all settings' figures are far closer to the actual.

So using the all settings figures, I believe that today's total fatality announcement of 28,489 is an actual total death figure of > 37,485.

Morning, squire.

I did two calculations after the numbers bounced up but haven't yet done today's numbers. As my statistic is the change in the ratio of new cases to deaths I expected a short blip depending on how the numbers were ajdusted, but once the adjustment has been competed (it will take a few days for this to finalise as first total deaths then total numbers of cases are recalculated, and I'm not surewhat the are doing, and even can do, about the latter) the statistic will settle back on its previous trajectory. This is because it is the change in a ratio - it isn't strongly affected by wobbles in accuracy.

As for your other question, the numbers we had all along are curated in two separate places, one that updates daily (but variably at the weekend) and the other that updates weekly. The weekly update has additionally a delay in it so it doesn't map entirely to the daliy data, and the weekly data include only deaths, not new cases as far as I can see.

Apologies if this all sounds a bit crap, and thinking about it, it is. I guess I'm not surprised because I know how some types of medical data are assembled. For example, in my research area, I asked a simple question 30 years ago: how many patients with a certain condition are give drug X, drug Y or drug Z. It took me ten years to finally realise that these drugs, used primarily in hospital, are chosen by the local committee of consultants, and this varies from hospital to hospital, and no central records of numbers and usage are kept. Thus, for an infectious disease you'd need every hospital, every GP practice (the two cases on my street were diagnosed down the phone by a nurse practitioner based on the patients' descriptions of symptoms only) to keep accurate records every time they diagnose a case, and every time they record a cause of death. So it isn't about having numbers all along; I think the government decided first to select data sources that are updated daily so a sense of trend of change could be seen (sticking to a single rubric) but have now changed this after being put under pressure for apparently grossly under reporting total deaths by inadvertently (I am being kind, here) excluding all care home deaths. I think they should have kept the two data sets separate and explained why, but they are politicians and when interfacing with Joe Public the political considerations dictate.

You are absolutely correct though that sudden big changes in the rubric for calculating any of this changes with affect the data hugely, albeit I don't think the noise will have a long term effect on my statistic. We shall see....it will be obvious as the UK numbers will suddenly stand out from the rest of the civilised world's.

As others have said repeatledly total number of deaths is ostensibly the key ststistic for assessing trends, albeit the exact numbers are questionable. As others have said in most countries this number is going down. In the worst performing nations it is too early to say. Those nations are Brazil, USA and, er, UK.

 

[Harry Wilson's tackle]

Latest number crunch. UK's rubric reboot has not had any lasting effect on trends. Numbers looking encouraging and fit with apparent global decline in new deaths. It seems that Iran, a nation from the lands that gave us algebra, are much more careful than Russia and Ecuador when making up their fake numbers, judging by their mad lurches in my statistic.

 

[Triggaaar]

As others have said repeatledly total number of deaths is ostensibly the key ststistic for assessing trends, albeit the exact numbers are questionable. As others have said in most countries this number is going down. In the worst performing nations it is too early to say. Those nations are Brazil, USA and, er, UK.

I don't understand how you are deducing that the number of deaths in the UK is not going down?

 

[Harry Wilson's tackle]

I don't understand how you are deducing that the number of deaths in the UK is not going down?

From John Hopkins page 2 minutes ago: Number of cases of cases, and number of cases of deaths (they haven't labeled the Y axes with very clear words, but the numbers and the titles should be clear enough)




It all depends on what we mean by 'number of deaths going down'. Obviously you don't mean the actual number of deaths is going down because that would mean resurrection (and that is happening only in Spain and France - presumably because they are Catholics - as I noted yesterday). I presume you mean the increase in the number of deaths is going down. Well, that isn't the case either as the number of deaths increases each day. What I will agree with is the increase in the increase in number of deaths is going down. Probably.

The trouble with the English language is that at a certain point the number of words necessary to achieve precision and accuracy becomes beyond that which can be accomodated by the attention span of your average punter. Perhaps we can reduce this all to "are we there yet?". And the answer is no, no we aren't.

 

[Triggaaar]

From John Hopkins page 2 minutes ago: Number of cases of cases, and number of cases of deaths (they haven't labeled the Y axes with very clear words, but the numbers and the titles should be clear enough)

View attachment 123095
View attachment 123096

It all depends on what we mean by 'number of deaths going down'. Obviously you don't mean the actual number of deaths is going down because that would mean resurrection (and that is happening only in Spain and France - presumably because they are Catholics - as I noted yesterday). I presume you mean the increase in the number of deaths is going down. Well, that isn't the case either as the number of deaths increases each day. What I will agree with is the increase in the increase in number of deaths is going down. Probably.

Are you kidding me? When talking about the number of deaths falling, we're obviously talking about the number of deaths for a given time period. In this case, we're talking about per day. So the number of deaths in the UK per day (from CV), is now lower than it was a few weeks ago.

So why have you replied with graphs showing the number of cases? I didn't say anything about cases.

The trouble with the English language is that at a certain point the number of words necessary to achieve precision and accuracy becomes beyond that which can be accomodated by the attention span of your average punter.

Don't be silly. It's obvious what we're talking about.

Perhaps we can reduce this all to "are we there yet?".

No, first you can explain why you think the number of deaths per day is not lower than it was.

 

[Triggaaar]

From John Hopkins page 2 minutes ago: Number of cases of cases, and number of cases of deaths (they haven't labeled the Y axes with very clear words, but the numbers and the titles should be clear enough)

Ah, so where it says 'number of cases', it means number of deaths. Are the figures based on when the data is released, or when the deaths occurred, or what?

It's easier to just look at the data from the UK government, who state when the deaths occurred, and the number of deaths are falling.

 

[Harry Wilson's tackle]

Ah, so where it says 'number of cases', it means number of deaths. Are the figures based on when the data is released, or when the deaths occurred, or what?

It's easier to just look at the data from the UK government, who state when the deaths occurred, and the number of deaths are falling.

Indeed.

However 'the number of deaths are falling' isn't precise. The number is (or may be) falling, provided that you explain that you mean the number of deaths on consecutive days is falling.

If you set that aside for a moment, just eyeball the figure below. It is number of deaths on a cumulative basis, as of now. Yes, the Y axis says cases but it means cases of death. For there to be no new deaths, those lines need to plateau. Until they plateau there will be deaths each day. The UK and US look furthest away to a plateau. No country (apart from China - ho ho) has declared a day with no deaths since it started. A massive optimist might extrapolate the UK's line and conclude there will be no new Covid deaths in 30 days. Neither you nor I know whether this will come to pass or not. I am never a fan of extrapolation at the best of times (see my oft-posted Mark Twain paraphrase).

That said if you look at my statistic updates you will see even I have been optimistic about where we may be heading now. Crikey, we could be all back at the Amex in a month, with our buckets and spades in the boots of our cars ready for a splurge down Shoreham Beach after.

 

[Notters]

Today's graph.

 

[Triggaaar]

However 'the number of deaths are falling' isn't precise.

Hey, it's you that said it.

The number is (or may be) falling, provided that you explain that you mean the number of deaths on consecutive days is falling.

We all know that's what we're referring to. What else could we mean?

If you set that aside for a moment, just eyeball the figure below. It is number of deaths on a cumulative basis, as of now. Yes, the Y axis says cases but it means cases of death. For there to be no new deaths, those lines need to plateau.

Yes I'm aware of that.

Until they plateau there will be deaths each day. The UK and US look furthest away to a plateau.

So? My point is that you said that the number of deaths per day was going down in most countries, but it was too early to say that in the UK. You're clearly wrong. It is clear that the number of deaths per day in the UK is lower now than it was. So where is your evidence that the number of deaths per day has not fallen?

 

[Triggaaar]

Today's graph.

View attachment 123104

Blimey, it looks like the number of deaths per day is sky-rocketing!

 

[Harry Wilson's tackle]

Blimey, it looks like the number of deaths per day is sky-rocketing!

That's very odd - where is the blip from when all the care home deaths were added?

 

[Triggaaar]

That's very odd - where is the blip from when all the care home deaths were added?

They're on a different chart. This chart clearly states it's the number of deaths in hospital (which I'm sure you know, you're just teasing).

Now I know you're a genuine poster HW, so when that chart clearly shows the number of daily deaths are falling, why are you stating that it's not clear yet?

 

[Harry Wilson's tackle]

That's very odd - where is the blip from when all the care home deaths were added?

Edit - oh, that's in hospital deaths not total deaths......

 

[Harry Wilson's tackle]

They're on a different chart. This chart clearly states it's the number of deaths in hospital (which I'm sure you know, you're just teasing).

Now I know you're a genuine post HW, so when that chart clearly shows the number of daily deaths are falling, why are you stating that it's not clear yet?

Total deaths. What I'm saying is I am not sure the trend to a fall in the number of new deaths is going to sustain. If it doesn't sustain we are not out of the woods. Like I said I am reluctant to extrapolate, and I wouldn't assume that in hospital deaths necessarily shows the true picture. But....it probably does, so, great - we may be home and hosed. We shall see.

 

[Harry Wilson's tackle]

Blimey, it looks like the number of deaths per day is sky-rocketing!

I would also add that if the number of actual deaths (as opposed to just in-hospital deaths) is plummeting, then if the number of new cases looks more 'steady' (looks pretty steady to me, first figure below) then the ratio of new cases to deaths should have already started to skyrocket from the lowly single figure numbers of a couple of weeks ago. But, hang on, they are still in single figures. Surely they should be up to Germany levels by now, or even better? April 29: Germany 25, UK 7. Today Germany 24 UK, er, 7. This suggests to me the in hospital death numbers are a big underestimate of total. Remember the steady number of new cases in the UK (see below) is supposed to be due to the massive increase in testing that has been done, and the high proportion of cases that are mild and nonlethal but cases nevertheless, comnfirmed by the new testing.

 

[Triggaaar]

Total deaths.

Total deaths what?

What I'm saying is I am not sure the trend to a fall in the number of new deaths is going to sustain. If it doesn't sustain we are not out of the woods.

None of us are sure the lower rate of deaths are going to sustain, but that applies to other countries too. I'm sorry if it feels like I'm picking on you HW, but this is what you said:

As others have said repeatedly total number of deaths is ostensibly the key statistic for assessing trends, albeit the exact numbers are questionable. As others have said in most countries this number is going down. In the worst performing nations it is too early to say. Those nations are Brazil, USA and, er, UK.

So you said the number was going down in other countries, but that we're one of the worst performing nations and it's too early to say. That's nonsense.

 

[Harry Wilson's tackle]

Total deaths what?

None of us are sure the lower rate of deaths are going to sustain, but that applies to other countries too. I'm sorry if it feels like I'm picking on you HW, but this is what you said:

So you said the number was going down in other countries, but that we're one of the worst performing nations and it's too early to say. That's nonsense.

No worries, mate. I stand by all your second to last sentence paraphrasing some of what I said or implied (albeit 'number going down' needs some explanation - cases or deaths? No, let's leave it...); certainly it looks like we are one of the worst performing nations, with new cases still hovering at a steady-ish level, and one of the smallest cases to deaths ratios, two indicators that are currently both shared only with America and Brazil. And yes, I still think it is too early to say (with high confidence) what is going to happen. That's why we don't have a start date yet for schools or football. And why my employer has already binned off all contact teaching that was supposed to start in October, with online teaching till January 2021 (that's a big London uni, too). Uncertainty is the problem. And I remain uncertain, guessing, doing back of envelope calculations, not dismissing all published data as false but not taking it all on face value. Getting my words wrong sometimes too. So it goes.

I don't mind you picking on my comments. It's easy to type 'cases' when one means 'deaths', and post reported hospital deaths as a surrogate for actual total UK deaths. Only by forensic analysis can we get to a position where we can hazard a wild guess about what might happen, given a fair wind

 

[Triggaaar]

I would also add that if the number of actual deaths (as opposed to just in-hospital deaths) is plummeting, then if the number of new cases looks more 'steady' (looks pretty steady to me, first figure below) then the ratio of new cases to deaths should have already started to skyrocket from the lowly single figure numbers of a couple of weeks ago. But, hang on, they are still in single figures. Surely they should be up to Germany levels by now, or even better?

You are getting yourself all in a muddle. The pertinent question is what is happening to the number of deaths per day - is it going up, remaining the same, or going down? The answer is that it's clearly going down.

That's all you need to know. So why confuse things by adding a spurious ratio?

April 29: Germany 25, UK 7. Today Germany 24 UK, er, 7. This suggests to me the in hospital death numbers are a big underestimate of total.

The number of hospital deaths is a reliable figure we can use. Even if total deaths were double that, it would still be a reliable metric. Conditions that would make it unreliable would include if people with the virus were originally able to go to hospital, but now they're not. This would arbitrarily keep the number of deaths in hospital down. But that is not happening. People are just as able to get hospital treatment now as before.

Remember the steady number of new cases in the UK (see below) is supposed to be due to the massive increase in testing that has been done, and the high proportion of cases that are mild and nonlethal but cases nevertheless, comnfirmed by the new testing.

What has that got to do with anything? The number of daily deaths are coming down. So what is your point about your ratio and testing?

 

[Triggaaar]

certainly it looks like we are one of the worst performing nations, with new cases still hovering at a steady-ish level

But that's because we're doing more testing! We're testing 10 times as many people as we were, so of course it's going to show more cases. If you just looked at the number of cases of each country a few weeks back, it would have looked like Germany were doing badly. Of course they weren't, they were doing extremely well, but they were testing a lot more people than us (and more than most other countries too). It's good that we're now testing more people, but don't then look at the comparatively higher total and think it's bad.

and one of the smallest cases to deaths ratios, two indicators that are currently both shared only with America and Brazil.

Our cases to deaths ratio is similar to many countries, like Italy, France, Spain, Belgium, Netherlands, Sweden. Adding in the fact that we're now recording more cases, because we're testing more, is completely misleading you.

And yes, I still think it is too early to say (with high confidence) what is going to happen.

I've not disagreed with that.

 

[Harry Wilson's tackle]

You are getting yourself all in a muddle. The pertinent question is what is happening to the number of deaths per day - is it going up, remaining the same, or going down? The answer is that it's clearly going down.

That's all you need to know. So why confuse things by adding a spurious ratio?

The number of hospital deaths is a reliable figure we can use. Even if total deaths were double that, it would still be a reliable metric. Conditions that would make it unreliable would include if people with the virus were originally able to go to hospital, but now they're not. This would arbitrarily keep the number of deaths in hospital down. But that is not happening. People are just as able to get hospital treatment now as before.

What has that got to do with anything? The number of daily deaths are coming down. So what is your point about your ratio and testing?

OK. The number of daily deaths in hospital is coming down. When you started criticising my comments I had never said anything about the number of hospital-only reported daily deaths. I am more than happy for you to show me that figure. I have explained why it tells only part of the story.

I also agree that of all the data published the figure you posted is probably the least equivocal. But it is not the number of daily deaths. People, you may be astonished to learn, die outside of hospital, too, at home and in nursing homes.