The Mad Modellers of Lockdown
19th May 2020
It appears we went into lockdown based on the modelling of one man – and his team. Neil Ferguson from Imperial College London. His workings predicted that, if nothing were done to prevent the spread of COVID, half a million people would die in the UK.
His prediction shaped the response of many countries around the world, definitely in the UK and the US. So, where did this half a million-figure come from? On a related note, the two million figure for the US is something which makes no sense at all.
This is because the US has five times the population of the UK. Thus, everything else being equal, in the US number should be two point five million. Even I can multiply 500,000 by five.
Getting back to Ferguson, and his model. So far, he has refused to release the data underpinning his model. Which, considering the impact it has had, is completely unacceptable. I think I would have given him a Chinese burn, at the very least, to get him to show me how he worked things out.
In truth, it is not exactly difficult to establish where this number came from. You can simply work backwards. There are sixty-six million six hundred thousand people living in the UK. If five hundred thousand die, that represents an infection fatality rate (IFR) of 0.75%. In other words, for every thousand people getting infected with COVID, seven and a half will die – on average.
Of course, there is an assumption built into the model that not everyone will get infected. Which is reasonable. There has been no pandemic in the history of the world where a bug managed to infect everyone – although it might be interesting to know why some people do not get infected, ever, when everyone around them is… This, I find, is the sort of question that never gets much looked at. Oh well.
Anyway, the Ferguson model predicts that eighty per cent of people could end up infected with COVID (which seems extraordinarily high and is simply a guess). That eighty per cent would happen if we all mingle and go to the pub, football matches, and suchlike. This increases the infection fatality rate (IFR) to 0.937% (0.75 ÷ 0.8). An IFR of 0.937% means that for every thousand people who get infected, nine and a half will die – on average.
The Ferguson team came up with an IFR of 0.9% (range 0.4 – 1.4), but I have no idea why it is not 0.937%. They talked about ‘mitigation’, but that didn’t seem to mean anything – it was just a fudge factor. Maybe they thought giving such a precise figure would look ridiculous when there are so many unknown variables flying about. True, but then again, I think the figures of 80% and 0.9 are simply wild guesses and look equally ridiculous.
The entire model can be seen in the original Ferguson model 1. By the way, I think I should mention that this paper was published on the 16th of March. Bear that date in mind.
So, that’s the model. Not very difficult really. Even though it is presented as some hugely complex mathematical monster, requiring the use of several super-computers running day and night to deal with the vast swathes of equations and data. Not so. You just need to do this:
66,600,000 x 0.009 x 0.8 = 500,000 (actually 479,520)
“Difficulty is a coin which the learned conjure with so as not to reveal the vanity of their studies and which human stupidity is keen to accept in payment.” Michel de Montaigne
You may wonder what the difference is between Case Fatality Rate (CFR), which is often mentioned, and the Infection Fatality Rate (IFR) – which is rarely mentioned. At the moment the case fatality rate (CFR) in the UK is well over 10%. This is clearly much higher than any predicted IFR.
The reason for this massive difference is because, if you only test people who are very ill, who have arrived in hospital (the bad cases of COVID), you are only testing those who are most ill, and most likely to die. Which gives you this very high CFR.
During any epidemic the CFR will always be high at the start, then start to fall, as more and more people with milder and milder symptoms are tested. Or, you later find out how many were actually infected.
However, unless you test everyone in the community, even those with no symptoms, the CFR will always be larger than the IFR. I hope this is clear.
This is a long-winded way of saying that no-one had much of a clue what the COVID-19 IFR may be. In the UK this is still the case, because no-one has a clue as to how many people have actually been infected.
All is not lost though, you can try to make a best guess, and you can do this by looking at the population, or country, where the greatest percentage of the population has been tested. At present that country is Iceland, total population 366,130.
With regard to the CFR in Iceland, as of the 10th of May, fifty-four thousand tests had been done. There were 1,800 positive cases, and the total number of deaths was ten, with no deaths for the previous three weeks. This represents a case fatality rate of 0.55%.2
This figure is the absolute maximum CFR, because it has not changed since the 19th of April, and there were another twelve thousand tests during that time, with only twenty-two more positive cases.
What does this tell us about the IFR? Well, we know that IFR will always be lower than the CFR. However, even if we assume that the CFR and IFR in Iceland are the same (which is next to impossible) the maximum death rate, in the UK, based on those figures, would be
66,600,000 x 0.0055 x 0.8 = 293,600
As an aside in Iceland they randomly tested 848 children and found that the number infected was 0.00%. Some of those children must have been exposed to the virus, so viral exposure clearly does not even, always, lead to asymptomatic disease…
The 0.8 figure (80% of the population getting infected) still seems extraordinarily high to me, but I am willing to let it go. Even though it looks that the total number of people who may become infected is almost certainly far, far, less than 80%.
Leaving that issue aside, what is the next step in analysing the figures. It is to add in the fact that, at least, fifty per cent of people who become infected with Covid-19 are asymptomatic. So, using Iceland, the IFR can only be a half of the CFR. Which gives us this figure for the UK
66,600,000 x 0.0055 x 0.8 x 0.5 = 146,800
Anyway, as of today, that figure is a pretty reasonable estimate of the absolute maximum deaths we could have seen, in the UK, had we done nothing. One quarter of the Ferguson number.
Has the Ferguson number changed? Well, it has certainly wobbled about all over the place. On the 5th of April, Neil Ferguson made this prediction
‘LONDON (Reuters) – UK deaths from the coronavirus could rise to between about 7,000 and 20,000 under measures taken to slow the spread of the virus, Neil Ferguson, a professor at Imperial College in London who has helped shape the government’s response, said on Sunday.’ 4
As I write this, we have had just over 34,000 deaths under lockdown. So, not quite seven to twenty thousand. Undaunted, on the 28th of April Professor Ferguson changed his mind again, and then gave this warning:
‘100,000 could die of coronavirus this year if a gradual lockdown lift is implemented to just shield the elderly, warns epidemiologist Prof Neil Ferguson – as new analysis warns 60,000 are predicted to die by start of August.’ 3
“Five hundred thousand” changes to “seven to twenty thousand”, then becomes a “hundred thousand” or maybe “sixty thousand”. One two, miss a few, ninety-nine a hundred.
Yes, of course, we all know that Professor Ferguson was recently found to have been, repeatedly, visited by a young married lady. Thus, flouting the very lockdown rules that he had done so much to create. The words delicious, and irony, spring to mind. That, however enjoyable it may be as a Shakespearean tale, of a man laid low by hubris, is not the main point.
The main point is why the bloody hell, how the bloody hell, did this man – and his group – come to hold so much sway. His figures underpinning the original model could not be verified, because he would not release the source data. Even if the figures had been available for scrutiny they kept swinging wildly about the place and have already been proven to be blindingly inaccurate.
The IFR of 0.9% is clearly, quite clearly, wrong. It is at least four times too high. The truth is that you could have given me a fag packet and a pencil, and I could have given you a more accurate model. Or we could have used Paul the Octopus whom you may, or may not remember:
‘Paul the Octopus (26 January 2008 – 26 October 2010) was a common octopus used to predict the results of association football matches. Accurate predictions in the 2010 World Cup brought him worldwide attention as an animal oracle.’5
Instead, our Government just kept repeating the mantra. ‘We are being led by The Science.’ As if Science required a definite article. Here is ‘the science’, let me show it to you. Crikey, and here’s me, I thought science was a bunch of ideas, conjectures and hypotheses used to try and explain the physical world around us. Constantly under debate, always changing. Never certain.
But no, it turns out it is an actual thing. ‘The science’. Boris keeps it in number ten Downing Street, and they share a cup of tea in the afternoon, along with a few jammy dodgers. Luckily ‘The Science’ is immune to COVID, so social distancing is not required. The Science also, probably, moves in mysterious ways.
“The science moves in a mysterious way
Its wonders to perform
It plants its footsteps in the sea
And rides upon the storm”
To be a little more serious, what is the science that is leading them? Mathematical models? Models that change and swirl and have little basis in reality. Models used to create predictions. As a friend has remarked to me many times: “there are two types of prediction – lucky and lousy”.
Our lives, our economy, our health service, all those people no longer getting treatment for other conditions, the heart attack patients not turning up at hospital, the cancelled cancer treatments, thousands of small businesses sacrificed at the altar of a mathematical model created by the mad modellers of the lockdown. Our lives, in their hands.
K’inell. As they say.