COVID-19: Day 108: Still concerned

April 18, 2020

Warning: Discussing death is difficult, and if you feel you will be offended by this discussion, please don’t read any further.

========================================

This is a short update from my previous posts (12, 3).

Slide3

Day 108 of 2020

On April 11th (Day 101) I correctly predicted that there would be no fall in COVID-19 hospital deaths this week just past.

My reasoning was that mortality from COVID-19 hospital admissions is around 20% and so 20% of new ‘Cases’ become ‘Deaths’ after 6 days on average.

The gist of this post is that the flat trend in COVID-19 cases has continued today, and that means that – to the extent the correlation I have observed continues to hold – the flat trend in deaths will continue for the next 6 days at least.

This means that by day 114 (April 24th) the death toll will exceed 20,000 – and we are still weeks away from substantially reducing new cases and getting back to work.

Given where the UK started – with a few days notice compared to other European countries – this represents a very bad outcome.

As some correspondents on my previous post pointed out, we urgently need to find the source of these ongoing infections and stop them.

Updated graphs

Slide1

The above graph shows various statistics plotted versus the day of the year.

  • the blue curve shows the daily published number of COVID-19 cases.
  • the red curve shows the daily number of COVID-19 deaths in hospital.
  • the black dotted line shows the predicted number of deaths based on 20% case mortality after 6 days.

Slide2

The above graph shows various statistics plotted versus the day of the year.

  • the blue curve shows the running total of COVID-19 cases.
  • the red curve shows the running total COVID-19 deaths in hospital.
  • the black dotted line shows the predicted number of deaths based on 20% case mortality after 6 days.

===========================

Discussing death is difficult, and if you have been offended by this discussion, I apologise. The reason I have written this is that I feel it is important that we all try to understand what is happening.

COVID-19: Day 107: I am concerned

April 17, 2020

Warning: Discussing death is difficult, and if you feel you will be offended by this discussion, please don’t read any further.

========================================

In the last couple of posts (1, 2), I have explained that it is possible to predict the number of people who will die from COVID-19 in a week’s time by looking at the number of cases confirmed today.

Day 107 of 2020

Slide1

The above graph shows various statistics plotted versus the day of the year.

  • the blue curve shows the daily published number of COVID-19 cases.
  • the red curve shows the daily number of COVID-19 deaths in hospital.
  • the black dotted line shows the predicted number of deaths based on 20% case mortality after 6 days.

With more data, I have revised my estimate of mortality (deaths ÷ cases) down from 25% to 20%, but the estimated time-to-death has shortened from 7 days to 6 days.

As the graph above shows, there has been no fall in the number of cases diagnosed and so the last 4 weeks of data lead us to expect that there will be no fall in the death rate – over 800 people each day – for at least another week.

Concerning

This is very concerning and indicates that whatever we are doing now is failing to eliminate the virus from circulation.

If I were in charge – I would want to know why the number of cases is not falling. If I didn’t know, I would recommend even more stringent lock down measures.

Why? Because by day 130, (4th May) I think our collective tolerance and forbearance will become severely strained. If the end is not in sight, and if that curve remains flat, then as the combined costs  (economic, social, personal, and medical) grow, I fear there may be social unrest and an already appalling situation will become uncontrollable.

The final toll

Slide2

The above graph shows various statistics plotted versus the day of the year.

  • the blue curve shows the running total of COVID-19 cases.
  • the red curve shows the running total COVID-19 deaths in hospital.
  • the black dotted line shows the predicted number of deaths based on 20% case mortality after 6 days.

At day 107, there have been 14,576 confirmed UK COVID-19 deaths. If today’s cases become death statistics in 6 days as they have for the last 4 weeks, then the total number of deaths will exceed 20,000 before the daily death rate has even begun to fall.

Earlier on in the crisis, it looked like the death toll could be kept well under 20,000. But that now looks impossible.

Until the number of daily cases begins to fall, it will be impossible to estimate how long our current ‘lockdown’ will need to last, and how great a cost we will all have to pay.

=============================

As I mentioned, discussing death is difficult, and if you have been offended by this discussion, I apologise. The reason I have written this is that I feel it is important that we all try to understand what is happening.

COVID-19 Hospital Mortality

April 12, 2020

Warning: Discussing death is difficult, and if you feel you will be offended by this discussion, please don’t read any further.

========================================

Yesterday, I concluded that the mortality of UK COVID-19  patients entering hospital i.e. people already seriously ill with COVID-19, was roughly 25%. I was shocked at this large figure.

Sadly, after further investigation it appears to be increasingly plausible.

UK hospital deaths versus age

UK Government data is available in a spreadsheet downloadable from a link on the COVID-19 ‘Dashboard’

The data show the age ranges of people who have died from COVID-19 in hospital. The age ranges are rather broad but I have taken the liberty of drawing a smooth line through the data points.Slide4Based on this data (and also shown on the graph) is my calculation of the average age of people dying in UK hospitals from COVID-19: it is approximately 74 years of age.

However this data does not tell us how many people in these age ranges were admitted to hospital, so we cannot calculate the mortality.

 

US hospital mortality versus age

The Washington Post has an article which includes data on COVID-19 mortality in US hospitals admissions versus age. The data is based on the admission of 6479 patients since 1st March 2020.  I have re-plotted the data below.

Slide5

This mortality is for US hospitals, rather than UK hospitals, but assuming that treatment is similar, then we can look at the expected mortality for patients at the average age of death of UK patients. This is shown below below with a red horizontal line indicating 25% mortality.

Slide6

This data seems self-consistent.

  • The relationship between UK daily cases and UK daily deaths that I discussed yesterday seems to indicate that mortality is around 25%.
  •  US mortality data shows that at the average of UK deaths, mortality for hospital admissions is 25%.

Ideally we would also like to know the ages of UK patients at admission, but I could not find that data.

Discussion

Notice that this only concerns patients who are admitted to hospital i.e. patients who are already poorly and who have generally been suffering at home. Most people recover at home without needing medical care.

But even so, I have again been saddened by this result which makes it less likely that yesterday’s analysis was in error.

This support gives increased confidence to the prediction  that the number of daily deaths for the next 7 days is unlikely to fall significantly, because these deaths correspond to people who have already been admitted to hospital.

=============================

As I mentioned, discussing death is difficult, and if you have been offended by this discussion, I apologise. The reason I have written this is that I feel it is important that we all try to understand what is happening.

COVID-19 Numerology

April 11, 2020

Warning: Discussing death is difficult, and if you feel you will be offended by this discussion, please don’t read any further.

========================================

Life is very pleasant for me and my wife in this ‘stay at home’ world, but I find myself permanently anxious and neurotically focused on ‘the numbers’: trying to understand them and use them to foresee what’s coming next.

I had thought naively that the ‘lock down’, which started on Day 81 of the year, would be completely effective, and that new cases of COVID-19 would begin to decline. But as the data below shows, that doesn’t seem to have happened.

Slide1

The number of new cases has stopped rising – but new cases are still occurring at around 4500 ± 500 cases per day.

As I understand the data, and the way in which testing is done, these are mainly people entering hospital. People who have probably been ill at home for some time, but their symptoms have now become serious enough for them to come to hospital.

But even so, some of those people will have been infected after Day 81.

Relating New Cases to Deaths

Some fraction of the people entering hospital will die a few days later.

I have looked at the UK data to try to understand how many people would die – the fractional mortality – and the delay.

To do this I took the ‘new cases‘ data and:

  • Applied a delay to the data that moves it to the right on the graph
  • Adjusted the fractional mortality to try to match the statistic for daily deaths. This moves it downwards on the graph.

Slide2

I found a reasonable match to the data for a delay of 7 days and a fractional mortality of 25%. i.e. the data seem to imply that 1 in 4 people being admitted to hospital as a new case will die, on average just 7 days later.

Slide3

Is this right?

Well obviously I don’t know if this is right or not.

I had expected a much lower mortality for people entering hospital – perhaps 1 in 10. On the graph above this would push the dotted black curve downwards.

But if that were so, then in order to match the ‘daily deaths’ data, the time to death would have to be very short, and in fact the curve doesn’t match the data well.

I found that reasonable matches could be obtained with:

  • mortality of 30%  and a time until death of around 9 days,
  • mortality of 20%  and a time until death of around 5 days,

But the best match (by eye) seemed to be with a mortality of 25%  and a time until death of around 7 days,

Discussion

I was shocked and saddened by this result. I hope I have missed something out or misinterpreted the data. Perhaps the mortality or time until death have improved throughout the last few weeks.

A mortality rate of 25% has been reported in the ‘worst hit’ hospitals, but I assumed this was exceptional. Also, the time until death seemed much faster than I had expected.

One additional feature of this analysis is that – if correct – it predicts the number of daily deaths for the next 7 days. And the prediction is disappointing.

The analysis indicates that the number of daily deaths in the next 7 days is unlikely to fall because these deaths correspond to people who have already been admitted to hospital.

Link to Excel Spreadsheet: Modelling Death Delay and Mortality

=============================

As I mentioned, discussing death is difficult, and if you have been offended by this discussion, I apologise. The reason I have written this is that I feel it is important that we all try understand what is happening.

 

UK Corona Virus Deaths in Context

April 3, 2020

Yesterday, April 2nd 2020, 684 people died from COVID-19 in UK hospitals.

But when we look back at 2020, we will see that the number of people dying in the UK was not significantly different from any other year. 

Really?

Yes Really. The graph below shows the number of recorded deaths in the UK from 1890 to 2018.

Slide2I was surprised by this number being so relatively constant (around 600,000) over this long period despite the doubling of the UK population (shown against the right-hand axis).

A reasonable estimate of the worst-case outcome of the pandemic, is that 20,000 people will die from COVID-19 in the UK this year.  I have shown the effect of this on the graph above. The detail shown in the green box is re-drawn below.

Slide1

The year-to-year variability over the period 2018 to 1990 is approximately ±11,000 deaths, and year-to-year variations in annual deaths of between 20,000 and 30,000 are common. See for example the 32,000 “extra deaths” in 2015 compared with 2014.

Fitting a trend line to the data from 2011 to 2018 we can estimate what would have been the likely number of deaths in 2020, and then see the effect of an additional 20,000 deaths. These are shown as two filled red dots.

So as I said at the start, when we look back at 2020, we will see that the number of people dying in the UK was not significantly different from any other year. 

But if that’s the case why all the fuss?

I am writing this because according to many media, the death of 20,000 people from COVID-19 represents a national failure. I disagree.

The actions we have already taken, and the tens of billions of pounds we have collectively spent, have already saved the lives of thousands – and probably tens of thousands – of people.

If we had not acted, then we would be looking at excess deaths on the order of hundreds of thousands of people –  as seen in the 1919 ‘Spanish Flu’ – or even more.

What next?

Based on my crude analysis, and assuming the “lock-down” has been very effective, then I expect the number of deaths per day to rise until approximately 11th April 2020.

The New York Times Tracker below shows that the UK’s ‘death curve’ is similar to Italy’s and so the final toll – which will not be reached until the end of April – looks like being close to, but slightly less than 20,000. Better outcomes are still possible.

Until then, I urge you to Keep Calm and Carry On.

CV Deaths in context

Data Sources

 

 

 

Life beyond lock-down: Masks for all?

April 3, 2020

Michael in a mask

Will we all be wearing masks in public for the next year or two?

A good friend sent me a link to a video which advocated the wearing of masks in public as a successful strategy for combating the transmission of corona virus.

I have no idea if this is true or not.

One thing of which I have been reminded by the current pandemic is that my intuition gained by experience as ‘an expert’ in one area, is not transferable. This pandemic has left me in a permanent state of bewilderment.

One of the key pieces of evidence offered in the video is the effectiveness of even primitive masks in inhibiting virus transmission in Czechia. Apparently,  mask-wearing in public has become de rigeur in Czechia, and there is – apparently – a low incidence of COVID-19 in Czechia.

I decided to look at the data

The table at the end of this article is compiled by data from Wikipedia’s list of the countries of Europe and their population, and the number of deaths recorded on Worldometer on the evening of 2nd April 2020.

The map below shows the results with the numbers expressing the numbers of deaths per million of the population.

[Note: Many European countries have small populations – less than the size of London – and many may not have good reporting of the deaths, which are in any case small in number. But the data is what it is.]

Number of deaths per million of population of countries in Europe on 2nd April 2020. See text for details. Data Table at the end of the article.

Number of deaths per million of population of countries in Europe on 2nd April 2020. See text for details. Data Table at the end of the article. Czechia is highlighted in yellow.

Does this data provide evidence that Czechia is a special case?

No.

To me it looks like Eastern Europe is generally less affected than Western Europe, and Czechia is in the middle. On the West it is bordered Germany and Austria, both of which have a low incidence (for Western Europe) per million of their population.

The 4 deaths per million of its population of its 10.7 million does not stand out as being anomalously low compared with, say, Poland (2 deaths per million of its population of 38 million) or Greece (5 deaths per million of its population of 10.4 million).

One further piece of evidence to look for would be the rate of growth of the virus within Czechia.

NYT Tracker for Czechia

The New York Times death tracker shows that the doubling-time for deaths in Czechia is similar to other countries in Western Europe – around 3 days.

The number of deaths are small and so the trend is uncertain, but it does not look like it is in the same group as Japan or South Korea which have only slow growth of virus-related deaths – a doubling time of more than 7 days.

In short, even though the idea of wearing a mask in public is not unreasonable, the data themselves do not seem to speak to the effectiveness of the habit.

But..

After the lock-down has ended, we all will need to be able to get out and about again and earn the money to pay for this hiatus. But the virus will still be out there and will still be exactly as lethal as it has been for these last few months.

So it might easily be that wearing a mask in public – proven in effectiveness or not –  may become a sign of respect for one’s fellow citizens.

One of the attractive features of the policy in Czechia is that the masks are not considered as being defensive i.e. protecting the wearer. Instead they are considered as a sign of pro-social behaviour i.e. a sign of one’s consideration of others.

Masks are unlikely to do any harm, and they may even do some good. But whichever is the case, it seems that in the US – the leader for many trends for both good and ill – their adoption may become mandatory.

NYT Tracker for Czechia

Headlines from papers on 2nd April 2020

So perhaps we will all be wearing masks in public for the for next year or two. I certainly didn’t see that coming!

UPDATE on 04/04/2020 ARTICLE ON ARS TECHNICA referencing new US CDC recommendations.

Data

Country Population Deaths @2/4/2020 Deaths/million
Germany 83,783,942 1,107 13
United Kingdom 67,886,011 2,921 43
France 65,273,511 5,387 83
Italy 60,461,826 13,915 230
Spain 46,754,778 10,348 221
Ukraine 43,733,762 22 1
Poland 37,846,611 57 2
Romania 19,237,691 115 6
Netherlands 17,134,872 1,339 78
Belgium 11,589,623 1,011 87
Czech Republic (Czechia) 10,708,981 44 4
Greece 10,423,054 53 5
Portugal 10,196,709 209 20
Sweden 10,099,265 308 30
Hungary 9,660,351 21 2
Belarus 9,449,323 4 0
Austria 9,006,398 158 18
Serbia 8,737,371 31 4
Switzerland 8,654,622 536 62
Bulgaria 6,948,445 10 1
Denmark 5,792,202 123 21
Finland 5,540,720 19 3
Slovakia 5,459,642 1 0
Norway 5,421,241 50 9
Ireland 4,937,786 98 20
Croatia 4,105,267 7 2
Moldova 4,033,963 6 1
Bosnia and Herzegovina 3,280,819 16 5
Albania 2,877,797 16 6
Lithuania 2,722,289 9 3
North Macedonia 2,083,374 11 5
Slovenia 2,078,938 17 8
Latvia 1,886,198 0 0
Estonia 1,326,535 11 8
Montenegro 628,066 2 3
Luxembourg 625,978 30 48

 

Keep Calm and Carry On

March 28, 2020

Friends,

Here is a piece of complete nonsense for you to contemplate during your self-isolation.

Best wishes

Michael

REFRAIN
Keep Calm and Carry On
Apparently, it’s how the war was won.
Well, I will agree to…,
… carry on with you
If you’ll agree to carry on with me

We all know the plain facts of the case.
Corona virus wants one per-cent of our race.
But we’ve collectively refused,
To be virally abused.
So, wash your hands and just carry on!

It’s all been so quick, it’s quite beyond belief.
And there’s little prospect of immediate relief.
But there’ll be toilet roll for all.
By the summer or the fall.
If you keep your distance and just carry on!

REFRAIN

After a month or two of isolation
We’ll be ready for Vcv celebrations
Some will have died to get us free
And we’ll drink to their memory
But now keep to yourselves and just carry on!

In the end we all see who the key workers are.
In factories and fields, and hospitals and bars.
They make life worth living for
Perhaps we’ll pay them more!
If we keep to ourselves and just carry on!

REFRAIN

Keep Calm and Carry on... with me

Keep Calm and Carry on… with me

Thoughts on UK corona virus

March 27, 2020

Well I didn’t see this coming!

And even now – as we all hunker down – I can scarcely believe it.

I struggled like many to make sense of the magnitude of what was happening, oscillating between a “this isn’t really very serious” and concern that “this really might be catastrophic“.

Eventually I understood that this was in fact very serious. I am writing this to try to express clearly my own understanding and explain why, slightly to my surprise, I am feeling positive about the UK’s response.

My sources of information have been:

In looking at the data I have ignored statistics about ‘cases’ or ‘hospitalisations’ because these terms are not clearly defined – and the true scale of the epidemic is still unknown. I have – grimly – only concentrated on deaths.

7000 UK lives saved so far

Deaths in the UK from Corona virus are shown below. The data were complete today 27th March 2020 and the national ‘lock down’ began on 22nd March, day 81 of the year. The deaths reported here would have been from people who were ill before the ‘lock down’.

UK CV deaths in contextThe vertical scale on the graph is logarithmic and can be tricky to read, but this type of graph allows one to  see the trends in data more clearly.

In particular – the rapid so-called exponential – growth typical of the early stages of an epidemic appears as a straight line, and the slope of the line tells us the time for the number of deaths to double.

If we extrapolate from data around day 74 – just two weeks ago – we might have plausibly expected many thousands of deaths by now. But looking at current data (day 85) we see the trend has significantly changed. So whatever we did – social distancing etc – has saved by my estimate around 7000 lives already.

I emphasise this because media coverage speaks only of failure and death. They make it seem like we are paying a great economic cost pointlessly. In fact, the price we are paying is high – but unless my understanding is flawed, we have already saved many lives.

I expect the initial three weeks of this current ‘lock down’ (out to day 102 in the year) will achieve more, but I think it will be extended, and there may a ‘super strict’ phase. The international lesson (see below) is that it takes around 6 weeks of the strictest measures to control the spread of the virus.

The above graphic is from the New York Times and is updated daily. It charts the number of deaths versus the time since the 25th death in that particular country.

(Link to New York Times Tracker)

What happens next?

Nobody knows. It’s the future and it is notoriously unpredictable, even guided by the experience of other countries (above).

But using the epidemic modeler we can make some guesses.

UK Epidemic Model

I took the basic numbers describing the spread of the virus from the internet (R0 = 2.4 etc.) and assumed there were 10 initial ‘seeds’ in the UK population of 66 million. I then chose the date of ‘lock down’ (Day 67 in the model but day 81 in the year) such that the death rate matched the observed death rate – about 50 deaths per day.

For a wide range of parameters, the model predicts that the peak in hospitalisations will occur in two to three weeks time – and there will be about 5 times as many patients as there are now.

It looks like the field hospital being built in the ExCel centre is likely to be required. Also around 10,000 intensive care beds may be required – more than twice those currently available (New Scientist Article) but this number is being increased rapidly.

  • So the model tells me to expect it to all get worse for about another two weeks. 

Of course a model is just a model, and people are not uniformly distributed around the country. In fact London, with about one eighth the population of the UK has about a third of cases. So the situation is likely to be worst in London.

So I expect that things will get worse. And the death toll will rise to probably between 10,000 and 20,000. But I feel it is important to bear in mind the actions we have taken will have saved hundreds of thousands of people from a premature death in the most appalling of circumstances.

Each year in the UK roughly 500,000 people die, and so – while each death is a loss – the likely death toll is … I can’t find the right word. But it is typical of the number of excess deaths from flu in a bad year.

Could we have acted sooner?

Technically, Yes. But – and I am not an apologist for this government – I think nobody could quite believe the scale of what was required, or the rate at which things were changing around the world.

By the time of the ‘lock down’, public opinion was overwhelmingly in favour of it, but even one week earlier that might not have been the case. If such an epidemic happens again, then I think our collective memory of the crisis will allow for much earlier action. This seems to be the case in countries in the Far East that successfully battled SARS and prevented it becoming a pandemic.

By the end of the year, our collective actions will have saved hundreds of thousands of lives. The epidemic modeler predicts that if the government had acted:

  • one week earlier, the death toll would have been ~5,000 rather than ~10,000.
  • one week later, the death toll would have been ~25,000 rather than ~10,000

In the longer term…

It is important to realise that even when the epidemic subsides – the corona virus will still be out there. When the social restrictions are lifted, the epidemic will grow again, until a vaccine is distributed.

Our experience of the crisis is likely to affect us collectively in too many ways to comment on here, but I am hopeful that this will have a positive effect both socially, and in our response to the climate crisis.

  • Air travel: This may never be the same. Ongoing quarantine regulations may yet mean that we might have already experienced peak-air. Wow!
  • Teleworking: Working from home and collaboratively across sites – organisations on the edge of adopting the technology will have been forced to do it. Vast numbers of commutes and business trips could be saved.
  • Cars: I am really enjoying quieter streets and I feel it will be hard to return the noise of even two weeks ago. Could the tide have turned against ‘excessive’ car use?
  • What government can do: We have seen that our government can act in dramatic ways with broad social support and – if we want it too – it can spend unbelievably large amounts of money on policies that people want, but which are not popular with many strong lobby groups.

Enough

I wish you all well wherever you are.

I am thinking in particular of colleagues in Italy and Spain where matters seem to be worse than in the UK, and of colleagues from China who have reached out to me with moving kindness.

Why no blog?

March 27, 2020

In case anyone cared or noticed

Sorry I haven’t been touch lately. I have been having a miserable time at work and found it hard to settle to other tasks requiring concentration. Like writing coherently.

I had hoped to work on for another year or two, but in the end, I just couldn’t bear the thought of it and I will now be leaving NPL at the end of April. With just 15 working days left I can feel my spirits rising.

And then hopefully my enthusiasm for writing will return.

Michael

 

 

Research into Nuclear Fusion is REALLY a waste of money.

December 4, 2019

In the previous post I argued that…

Research into Nuclear Fusion is a waste of money

I felt pleased with this article – writing it helped to clarify my thoughts.

In particular (although I didn’t  express this as clearly as I wanted to) I realised that it wasn’t just the specific overwhelming technological problems that made Fusion Research a bad idea. Any power generation scheme that is that complex and expensive will inevitably never be built. I illustrated this with a quote I heard some time ago.

I don’t want to live in a world with
nuclear fusion reactors, because
I don’t want to live in a world
where electricity is that expensive.
Unknown author

It was only after I had written the article that a colleague at work pointed out another article that said the same thing.

Why fusion will never happen

Distressingly, this article is much more clearly written.

It was published by Maury Markowitz in 2012, and amusingly illustrates Fusion Power with a picture of a unicorn jumping over a rainbow. 

art_trade__unicorn_and_rainbow_by_royalty_9-d2y1jq1

Fusion Power (from Matter 2 Energy)

Since then the fundamental truth has only become truer.

I feel a certain sadness in acknowledging that this long-hoped-for technology will never materialise.

But once understood, there is only one rational path of action: we should stop throwing good money after bad and stop funding fusion research right now. 

 

S


%d bloggers like this: