Re: Coronavirus - Numbers and Statistics

[Newroad]

Definitions, JohnHemming.

https://www.oxfordlearnersdictionaries.com/definition/english/virulence: "The degree to which a disease or poison is dangerous or harmful and quick to have an effect "
https://www.merriam-webster.com/dictionary/virulent: "Marked by a rapid, severe, and destructive course "
https://en.wikipedia.org/wiki/Virulence: "Virulence is a pathogen's or microbe's ability to cause damage to a host"

I could go on. The point is clear - unless the above is what you intended (and similar for "milder") you should explain your atypical use of such terms up front.

I prefer my own somewhat classic way of quoting - it's somewhat more succinct - and am not bothered if others wish to follow up or not. However, I can understand those who have a different preference.

Regards, Newroad

[tjh290633]

I understand that nowadays everyone who goes into hospital, for whatever reason, is tested for coronavirus, and many not displaying symptoms have been found to be infected. I believe that this was not the case during the "first wave", and so the reported numbers of "cases" then may be a vast underestimate.

Is the so-called "second wave" reflecting this effect? Might we be panicking needlessly? I await the Health Secretary's briefing with interest.

Incidentally I only know of one person who has tested positive, one of my grandaughters, who had no symptoms and all her domestic contacts tested negative.

TJH

[Newroad]

Hi TJH.

I hope it steers clear of you! I've had it, as did my wife and one of my children for certain (the other likely too, but asymptomatic): https://www.lemonfool.co.uk/viewtopic.php?f=98&t=27026

I agree comparison of Covid-19 statistics over any length of time is almost meaningless - changes in testing capacity and protocols etc. Over a shorter timeframe, comparisons are plausible. I wouldn't draw the same conclusion as you however.

As it happens, latest daily report from the UK is 50K+ cases and almost 1000 deaths*.

Regards, Newroad

* usual caveat - defined as having tested positive for Covid-19 in the preceding 28 days

[ReformedCharacter]

My OH who works at a school has received some bumf from the DOE regarding the Lateral Flow Tests, it says:

The Lateral Flow Tests are as accurate as the PCR tests (99.8%).

Yet an article in the BMJ says:

The rapid test kits most widely used in UK universities, schools, and care homes detect just 48.89% of covid-19 infections in asymptomatic people when compared with a polymerase chain reaction (PCR) test, real world data from the Liverpool pilot have shown.

The Innova Lateral Flow SARS-CoV-2 antigen test failed to detect three in 10 cases with the highest viral loads, in preliminary data released from the field evaluation of testing in asymptomatic people

https://www.bmj.com/content/371/bmj.m4848

Doesn't inspire confidence.

RC

[Mike4]

I understand that nowadays everyone who goes into hospital, for whatever reason, is tested for coronavirus, and many not displaying symptoms have been found to be infected. I believe that this was not the case during the "first wave", and so the reported numbers of "cases" then may be a vast underestimate.

Is the so-called "second wave" reflecting this effect? Might we be panicking needlessly? I await the Health Secretary's briefing with interest.

I'm inclined to trust the reports of hospitals overflowing with COVID admissions as an indicator of the true level of infections. So it appears to me infection in the population is probably higher now than during the first wave.

I expect Mr Hancock will say panicking now is the right thing to do, at the right time, guided by the science.

[johnhemming]

I'm inclined to trust the reports of hospitals overflowing with COVID admissions as an indicator of the true level of infections. So it appears to me infection in the population is probably higher now than during the first wave.

Numbers of new infections per day was clearly higher in the first wave. The seasonal wave, however, is flatter and there will have been more infections during it than during the first wave. However, this varies from place to place in the country as can be seen from the NHS trust statistics.

[dealtn]

I understand that nowadays everyone who goes into hospital, for whatever reason, is tested for coronavirus, and many not displaying symptoms have been found to be infected. I believe that this was not the case during the "first wave", and so the reported numbers of "cases" then may be a vast underestimate.

Is the so-called "second wave" reflecting this effect? Might we be panicking needlessly? I await the Health Secretary's briefing with interest.

I'm inclined to trust the reports of hospitals overflowing with COVID admissions as an indicator of the true level of infections. So it appears to me infection in the population is probably higher now than during the first wave.

Not necessarily. It isn't just the hospitalisations (as a result of infections), but the length of stay. Frankly some of those hospitalised in the first wave didn't need a bed for as long. Treatments have got better which ironically means beds are being used more (although ventilators possibly less).

[johnhemming]

Is the so-called "second wave" reflecting this effect? Might we be panicking needlessly? I await the Health Secretary's briefing with interest.

I think scotia's work highlights that about a quarter of the admissions will die (probably of covid, but not necessarily).

Incidentally admissions are now higher (2752), but not as high as the first wave (3099) ... yet.

Still really a South/East thing for growth although the South West had a jump.

[Newroad]

It's an interesting discussion Mike4, JohnHemming and Dealtn.

Instinctively, I would have said general level of infection in the populace was higher during the current wave than the first, but could easily be wrong - or it might be a regional consideration.

On balance, there are three somewhat different things being considered - recorded new daily infections, recorded new daily hospital admissions and net current hospital admissions. In the end, it is predominantly the 3rd of those that matters (in the sense of the NHS potentially being overwhelmed) but of course the first and the second are indicators with different lead times and correlation.

As Dealtn notes, we now have better knowledge and tools as to how to treat it (Dexamethasone in particular, but also body positioning and a reduced emphasis on using ventilators) and it may well be that fewer early patients survived, or survived as long.

Regards, Newroad

[scotia]

A word of caution on the statistics being quoted over the Christmas Period - particularly the deaths at the date of registration (not at the date of death). In normal times these vary substantially with the day of the week - so only make sense when averaged over 7 days. But the Christmas holiday has also produced a further distortion. So today (30/12/20) the recorded figure for England is 919 - and this was highlighted on the news as being near 1000 (for the UK?) . But the figure for 26/12/20 was only 175. So take all such numbers with a large pinch of salt. One-day numbers are meaningless. Even the weekly average is going to bounce about until the holiday effect disappears.

[johnhemming]

If you look at this
https://coronavirus.data.gov.uk/details/deaths
and click on data

it displays the figures by date of death.

[scotia]

If you look at this
https://coronavirus.data.gov.uk/details/deaths
and click on data

it displays the figures by date of death.

Yes - but unfortunately the date of death statistics have a considerable lag before they settle down to a final value. So they are only of use in examining semi-historical data.
They were my statistics of choice when I went back to examine the correlations between admissions and deaths around the first peak of the epidemic. However they proved useless - since in the first few weeks of April the excess deaths curve showed that Covid-19 related deaths were being significantly under-reported. And my attempted crude fit also confirmed this feature. So it was one lesson that I learned - don't make any projections on the published Covid-19 deaths data (by registration or by date of death) around the peak of the first epidemic. I should add that I don't put any blame on the doctors for this lack of accuracy - it was a panic situation, with all hands to the pumps, and accurate statistics on the cause of deaths would not be a high priority.

[jfgw]

Re:

https://johnhemming.blogspot.com/2020/1 ... -data.html

I have looked up the populations covered by 20 trusts and produced a scatter diagram:



As far as I can tell, it shows us knickers-all and I might as well have fired a shotgun at a piece of paper.

The increases and decreases don't seem any less random either, whether ordered by total admissions per million or Fortnight First Wave per million.

It might be worth looking at admissions per number of beds.

My conclusion so far is that looking at individual NHS trusts is either fruitless or requires a lot of things to be taken into consideration. I did try to avoid specialist trusts such as cancer hospitals, mental health units and Nightingale hospitals.


Julian F. G. W.

[johnhemming]

I have looked up the populations covered by 20 trusts and produced a scatter diagram:

I would be interested to know the source you used for this. My understanding of trusts is that getting this information is quite difficult.

If I were to be doing anything with these figures, however, I would look at the total admissions to date (per population) not the recent admissions.

[jfgw]

I have looked up the populations covered by 20 trusts and produced a scatter diagram:

I would be interested to know the source you used for this. My understanding of trusts is that getting this information is quite difficult.

Many are reported on the trusts' websites. Simply Googling usually brings up a figure in these cases. With some trusts, the data are not so readily available.

If I were to be doing anything with these figures, however, I would look at the total admissions to date (per population) not the recent admissions.

I was using your data to try to establish whether there was a link between the first wave admission rate and the more recent admission rate. I only had your table to go on as I don't have an easy means of downloading and extracting these data from the government site (although, as I am only listing a fairly small number of trusts, I could potentially download the data sets one at a time).

I have produced the following table based upon admissions data from your table and population data sourced as above:



There does appear to be some bias toward "Reducing" for the higher rates of total admission. I would want to rule out other potential reasons for this, however, such as patients being taken to hospitals with more available capacity. I also notice a bias toward larger populations for the lower rates of total admission which suggests that the population numbers obtained are not necessarily the most relevant. The number of beds may be more suitable, and this could alter the order. "Total number of beds" is unlikely to be ideal, however.

At present, I am dubious about the usefulness of individual trust data. I also think that I need to add more trusts to the list before drawing conclusions.


Julian F. G. W.

[johnhemming]

At present, I am dubious about the usefulness of individual trust data. I also think that I need to add more trusts to the list before drawing conclusions.

What you want to focus on is the acute trusts. The first on your list is a mental health trust. I simply treat them as small and ignore them. If there is a small number of admissions, but we should also ignore the larger ones. I assume the cases are ones where people have caught the infection in the trust rather than ones where people have been admitted because of covid.

The key question is what number of admissions per population are likely. That depends upon the demography, but a good starting position is just to look at the rate per population.

For areas such as Guys and Kings that should be a limiting figure, but there are some transfers between trusts particularly in today's figures.

[redsturgeon]

At present, I am dubious about the usefulness of individual trust data. I also think that I need to add more trusts to the list before drawing conclusions.

For areas such as Guys and Kings that should be a limiting figure, but there are some transfers between trusts particularly in today's figures.

I pointed this out earlier in the week and you tried to play it down.

John

[johnhemming]

I pointed this out earlier in the week and you tried to play it down.

I know it happens. It was not visible in the previous figures, but is visible in these ones. There are new figures so far released some time on Thursday afternoon each week.

And as I have said it does not affect the conclusion that so many people were infected in those areas that they hit the HIT not just for the summer, but also for the winter.

That, of course, is a key fact that can be used to make estimates as to the HIT measured in hospital admissions, but we don't have all the data for that. What we need is a realistic figure as to the population served by Guys, Kings and possibly Imperial. Imperial looks to be having some form of seasonal wave, but the other two look to be serving other hospitals more recently, but otherwise not having any material second wave.

[swill453]

I pointed this out earlier in the week and you tried to play it down.

I know it happens. It was not visible in the previous figures, but is visible in these ones. There are new figures so far released some time on Thursday afternoon each week.

And as I have said it does not affect the conclusion that so many people were infected in those areas that they hit the HIT not just for the summer, but also for the winter.

That, of course, is a key fact that can be used to make estimates as to the HIT measured in hospital admissions, but we don't have all the data for that. What we need is a realistic figure as to the population served by Guys, Kings and possibly Imperial. Imperial looks to be having some form of seasonal wave, but the other two look to be serving other hospitals more recently, but otherwise not having any material second wave.

There is, of course, no evidence that the HIT has been, er, hit anywhere.

(Other than tiny populations like individual households.)

Scott.

[servodude]

I pointed this out earlier in the week and you tried to play it down.

I know it happens. It was not visible in the previous figures, but is visible in these ones. There are new figures so far released some time on Thursday afternoon each week.

And as I have said it does not affect the conclusion that so many people were infected in those areas that they hit the HIT not just for the summer, but also for the winter.

That, of course, is a key fact that can be used to make estimates as to the HIT measured in hospital admissions, but we don't have all the data for that. What we need is a realistic figure as to the population served by Guys, Kings and possibly Imperial. Imperial looks to be having some form of seasonal wave, but the other two look to be serving other hospitals more recently, but otherwise not having any material second wave.

There is, of course, no evidence that the HIT has been, er, hit anywhere.

(Other than tiny populations like individual households.)

Scott.

Or possibly Manaus?

-sd