Skip to content

Thoughts on Coronavirus

Last updated on November 24, 2020

It was approaching midnight on December 31st 2019 when I said to my wife – “Isn’t it interesting how everyone talks at New Year about the year ahead, their plans, their expectations, their predictions. Yet every year something happens that comes completely out of the blue.”

Five months later I was sat talking to my wife about that very conversation. Having spent the previous 10 weeks inside our house, just leaving once a week to put rubbish in a dumpster behind our apartment block.

Had my wife suggested in December that we’d spend the entire spring effectively under house arrest I would have laughed until long after midnight.

I’m sure in the future historians are going to look back at 2020 the way we look back at the Second World War. The entire world shook-up by a single seemingly never ending event. Citizens’ movements and behaviours have been restricted in a way not seen before during peace-time. We’ve all lived through an event that will be talked about for centuries. I don’t think this has yet sunk in.

While I’ve never really stopped thinking about the pandemic since those early reports in January, I have been pondering some things in particular recently and wanted to note down some of my thoughts.

Data In Action

The idea that the world should be run by data rather than doctrine has been gaining popularity since the enlightenment. The computing revolution of the last 50 years has made this idea a workable reality. From the largest websites adjusting their content dynamically based on millions of recorded data points, to a random person deciding to go for a longer walk each day based on the number of steps their watch records, we’re all at it.

Pure accurately recorded data has the benefit of being neutral. The watch on your wrist recording your heart beat doesn’t care what mood you’re in. The thermometer on a weather station doesn’t care about climate change politiics. The vehicle counting device on a road outside a village doesn’t care about budgets for infrastructure projects.

So far so good, technology let’s us measure reality, and then make decisions based on those recordings and not a handful of random anecdotes or strong opinions.

Now fast forward to 2020, and the deadly pandemic we’re currently experiencing. What a perfect moment for data analysis to save lives and direct humanity towards a route back to normality.

We have a new disease (Covid-19), we have a highly contagious new virus (SARS-CoV-2), and we have new tests to detemine if a person has the virus inside them, or has had it previously. At the time of writing we have no readily available vaccine, so we need to stop people from getting it so they don’t get sick from it.

We don’t want people to die, and we don’t want them to suffer in hospital either. So we’re trying to modify our actions to limit those events as much as possible.

Deaths follow cases, so to limit deaths we just discover cases, and then limit the actions of those who have it to stop them passing it on and creating more cases. Easy right?

Theory vs Reality

The first sign that our proposed solution might not be so straightforward is the fact that testing somebody for coronavirus takes effort. A swab needs to be taken, which then needs to be analysed, the person then needs to be informed of the result.

The second problem is that as early as February 2020 it was clear that not everyone who has the SARS-CoV-2 virus inside their system realises it. Estimates vary but between 20% and 50% of all cases might well be asymptomatic, meaning without any recognisable symptoms.

The fact that testing someone for coronavirus takes manual effort means we can’t test everyone all the time, therefore we need to pick people to test. Now who do we test? People who come forward with symptoms, sure. But what about the people who have mild symptoms and don’t come forward and seek medical advice?

Estimates vary but the World Health Organisation in March 2020 estimated that at least 80% of cases don’t require hospitalisation. Even if we take that as a low estimate that’s a huge amount of people who will have Covid-19, and be contagious, but won’t be ill enough to call their doctor or hospital with serious medical issues. Even if many of those do voluntarily go in for tests, you can be sure a majority probably feel safer just staying at home with their mild symptoms.

Then of course there are those with no symptoms at all, who wouldn’t volunteer for a test because they have no idea they have it.

To summarise we need to stop people who have it from giving it to others, but we can’t because we don’t know who has it.

Test & Trace

To counteract the above countries have spent literally billions on ‘test and trace’ schemes. This is the idea that if person A is confirmed to have the virus, then by testing people who they have recently interacted with we can discover mild but contagious cases nearby and ensure those people isolate immediately, thus stopping them spreading the virus further.

Again, this is great on paper, but does it work in reality?

Let’s say person X is in hospital and a test has confirmed they are suffering from Covid-19.

They live with their wife and three teenage children.

A well funded ‘test and trace’ scheme armed with plenty of testing capacity firstly asks the wife and children to get tested, seeing as they live with the man in question. The tests for the wife and one of the children come back positive. Prior to the test they were asked to list people they had come in contact with recently. The wife lists her friend next door who lives on her own, the child lists four friends with whom they went to the cinema earlier in the week.

The friend next door gets tested and comes back positive, as does their sister and her husband who they had over for drinks the afternoon the hospitalised man’s wife came over. Two of the childrens’ friends come back positive too.

So that’s 8 confirmed positives from the original man in hospital. However – other than the woman next door feeling a bit rough one day, and one of the children who went to the cinema commenting that they have a strange cough, nobody else can recall any serious symptoms. The family isolates and everyone is fine, the official data records 8 cases of Covid-19.

Now let’s take the same man, but place him in an area with a less well funded test and trace system. He is confirmed as positive and his wife is asked how she feels, she says she is absolutely fine, the person responsible for ‘testing and tracing’ asks who the man lives with, the wife mentions the kids saying they’re fine with no reported issues. Nothing is done. The wife does take a test and the result comes back nearly 3 days later and is positive, the wife isolates at home and the husband eventually rejoins her. The children also isolate but are not asked to get tested. A couple of weeks later everyone is absolutely fine. The data records 2 cases of Covid-19.

In both of these cases the hospital workers are doing their jobs properly, the people responsible for ‘testing and tracing’ believe they are asking the right questions. Nobody dies. The official government records show that particular neighbourhood having 8 cases of Covid-19 in situation one, it records the exact same neighbourhood having 2 cases of Covid-19 in situation two.

Nobody is trying to mislead anyone here, it’s just the reality of trying to pinpoint cases of a contagious disease that is 99% non-fatal and mild in 80%+ of cases. The further you look the more you will find.

Neither of these scenarios include the man who came in to read the electricity meter and spoke to the wife, getting infected after touching a surface the wife coughed on moments before opening the door to him. He is ill for a week and only his stubborness stops him calling the hospital for an ambulance. His case is never recorded.

Data scientists have been very busy compiling data when instead they should have been yelling from the rooftops about how much missing data there is.

Absence of Evidence does not equal Evidence of Absence

You can tell with great accuracy what the temperature was at midday outside your house. You can tell with accuracy what your average heart rate was between the hours of 9am and 10am this morning. You cannot tell with accuracy how many cases of Covid-19 there are in your country. You can only report a reasonably accurate absolute minimum figure.

Case figures then become more a measure of how good you are at testing than how prevalent the virus is.

Now it’s not that the case numbers aren’t of importance, of course they are. When test and trace schemes are consistent, and testing mechanisms are reliable and well timed, the case numbers in a particular place one month to the next can give a solid indiciation of whether the virus is spreading rapidly through the population or not.

In reality though people in the UK spent February and the first two weeks of March reading about every single new case of Coronavirus that was reported. On February 6th a third case was reported, on February 10th the total stood at 8, then on February 29th the first person without a history of recent overseas travel tested positive.

We were all carefully following a narrative about a tiny number of cases gradually building. Yet, the reality was clearly a whole different story. On March 5th the official number of cases had reached 115. If we work on the assumption that roughly 1% of those with the virus end up dying, that would suggest perhaps 1 or 2 deaths were likely a couple of weeks later.

Let’s look at the official UK Coronavirus website to see the number of deaths later in March. (https://coronavirus.data.gov.uk/deaths)

On March 31st 2020 603 people in England died after having had a test showing they were infected with Coronavirus.

This Telegraph article from March 12th 2020 reports that Coronavirus kills in approximately 18 days. So we can assume those 603 people probably caught it on or around March 13th.

(https://www.telegraph.co.uk/news/2020/03/12/coronavirus-kills-average-185-days)

Now catching Coronavirus isn’t like being hit by a bus, when it happens you have no idea it just happened. According to the World Health Organisation it takes on average 5-6 days for symptoms to show up. Let’s be fair and assume those who died from it on March 31 were particularly vulnerable to it, so their symptoms showed on day 4. They would then either call a doctor, or go into hospital, and then be tested for the virus. Their positive result would then come anything from a few hours to a couple of days later.

Let’s take a step back and analyse this

As of March 5th 2020 in England just 115 people in total had tested positive for Coronavirus. Many thousands of negative tests had been returned, confirming that they were activitely looking for cases and not finding them. Most cases were linked to other cases, creating a belief in the UK that the virus was under control and normal life could continue. Large sports events took place with many thousands of people attending.

On March 31st 2020 in England – 603 people died on that day alone after definitely suffering from Covid-19

Those 603 people caught it about 18 days earlier, so that’s March 13th.

But their symptons didn’t show for 4-5 days, when symptoms arrived they got medical attention and a test, which then puts it at the earliest 5-6 days after infection. So that’s March 19th.

On March 19th 2020 there were 930 confirmed cases of Coronavirus discovered in England.

The 603 who died on March 31st should be in that cohort.

Now estimates of the mortality rate for Covid-19 have been changing on a daily basis.

Assuming a 1% mortality rate (on the high side, according to recent reports). The 603 dead would be the 1%, with the other 99% being the 59,697 that together make up 60,300 cases.

This tells us that approximately 60,300 people caught Coronavirus on March 13th in England. The official figures list 930 new cases. The world had been on high alert since January with near constant news updates on the events firstly unfolding in China and then elsewhere as the virus spread.

In the UK newspapers ran headline news when new cases were discovered throughout February and early March. (https://en.wikipedia.org/wiki/Timeline_of_the_COVID-19_pandemic_in_England), it was already over a month since on February 11 the 9th case was confirmed in London.

So even with the whole world on high alert, the number of cases in England was being under reported by at least a factor of 60.

What this tells you is that “case data” from the early stages of the pandemic is absolute junk. Yet we still see the March/April graph of cases in nearly every government powerpoint presentation. That data has been taken as fact since the day it was released. Governments probably need to discard case data from before June when it comes to determining public policy.

Now hopefully you’re still reading, I appreciate that most people understand that case numbers in the early stages were under estimated.

What we’ve shown is that test and trace schemes while a good idea in practice, in reality have varying degrees of effectiveness, and have a strong habit of finding more cases when given more resources to look for them.

We’ve also seen that even when a country is on high alert and the narrative is almost ‘searching for cases’, tens of thousands of cases got missed in the first three weeks of March.

The typically mild nature of the Covid-19 sickness and confirmed fact that many cases are entirely lacking in symptoms means that not even a test and tracing scheme with a budget of billions of pounds in the UK can possibly find even the majority of cases.

Yet we still read a narrative that is obsessed with the exact number of cases each day for a particular area. This happens where I live. One day it’s say 1203 cases, then the next it’s 1434 cases, then gasp, it’s 1384 cases and I read people saying “the tide has turned!”.

At worst case numbers are almost meaningless, at best when combined with consistent testing they can hint at whether local numbers are on the way up or on the way down, and whether that rate of change is fast or slow.

The Elephant In The Room

At this point you’re probably sick to death of case numbers, so I’ll keep this brief. Not all cases of Covid-19 are the same. From early on it became very obvious that older people are far more likely to die from it than teenagers.

Here is a table released by the Center For Disease Control in the USA (CDC). This is a governmental organisation with a budget of around $7 billion USD a year, so we can assume they know what they’re talking about.

(https://www.cdc.gov/coronavirus/2019-ncov/covid-data/investigations-discovery/hospitalization-death-by-age.html)

My wife is 29, so she fits into the 18-29 category on that table. I believe our neighbour is roughly 65 years old, and appears healthy. Therefore all other things being equal our neighbour is 90 times more likely to die from Covid-19 than my wife.

Even ignoring deaths for a moment, and just taking into account hospitalisations. Our otherwise healthy neighbour is still 5 times more likely to need to go into hospital than my wife.

What this means is that case numbers alone are nearly useless when determining if a region can cope with a sudden increase of Covid-19 cases.

Take one example, a nursing home reveals 55 positive cases for Coronavirus, I read this in the local paper the other day. Those nursing home patients are likely to be in the 75-84 age category. This means they are at least 220 times more likely to die than someone in the 18-29 age group.

Let’s say elsewhere in the city there are 55 positive cases announced connected to a university student union event. Those people will be overwhelmingly in the 18-29 age group.

A death from Covid-19 is a tragedy in itself, but it also involves a patient spending typically a week or two in a hospital bed. The resources involved in trying to help that patient are significant, and involve doctors, oxygen tanks and many tests and other medicines.

If we assume a generous on the low-side 5% death rate for those in the 75-84 age group (which could easily be higher as those in the nursing home already don’t include people fit enough to still live by themselves), then that translates into roughly 3 deaths from the 55 cases. (In reality it could be more as clearly a nursing home could be home to even older residents, the 85+ age group is 630 times more likely to die than the 18-29 age group).

Going back to the 55 university students – to get the predicted 3 deaths from that cohort you would need 220 times as many positive tests, given the reduced death risk for people in that age group. So those 55 positive tests would in reality need to be 12100 positive tests to become the same problem for society.

Therefore when talking about what we’re trying ultimately to avoid, death, the underlying people behind the case numbers are critical to know. Hearing of “55 new cases” is meaningless in itself without knowing what type of people we’re talking about.

Leaving Cases Behind – For A Better Measure

So to summarise, we cannot properly measure cases, we find more cases when we look for them and case numbers alone cannot possibly predict future death numbers.

It seems to me that recording the number of people dying having confirmed symptoms of Covid-19 and typically a positive test result as well is a measure that we can make. We can then monitor that number to determine if the virus is causing a problem to society or not. The problem is that should that number start to rise a lot, then any action we take to minimalise it will have a lag effect of 15-30 days at least.

Measuring when somebody is assigned a hospital bed and has had either a positive test or strong symptoms of Covid-19 is also doable and accurate. This data point when assessed on a hospital by hospital basis should give a good prediction of the number of deaths from Covid-19 at that particular hospital in the 7-14 days following the admission of each patient. You would need to do this on a hospital by hospital basis as health outcomes are not always equal in different areas of a country.

By combining the figures given for each hospital this would produce a death prediction for the 7-14 days ahead, and is the only reliable predictor of Covid-19 fatalities we have.

Trying to improve the predictive process by bringing case numbers in is in my opinion a jump from solid data to wildly unpredictable data. A surge in cases in a university town isn’t actually a huge problem, whereas a surge in cases in a coastal town popular with retirees could be a tragedy about to happen.

The Resource Management Falacy

Finally, just when we think we’re back on solid ground when it comes to data analysis, there’s another problem to consider.

It’s no secret that hospitals have become a lot better at treating Covid-19 as the months have passed. Early on ventilators were seen as the solution, whereas now they are seen as an absolute last resort, with pure oxygen being a far better aid for severe Covid-19 cases.

Doctors and nurses are learning what works and what doesn’t work. Many drugs have been tried and tested, with some proving very useful when provided at the right time. The scientific method has allowed us to test ideas and record what works and what doesn’t work, with many thousands of severe cases out there the data is solid enough to show what definitely helps and what definitely has minimal effect.

So far so good, but any hospital treatment requires doctors and nurses. A doctor and her team might have 10 cases of severe Covid-19 that they’re monitoring. They’re able to do their best but they still lose a patient. Two of the patients are helped using a new intensive therapy the doctor heard about from colleagues at a different hospital, it took careful effort and monitoring but they were able to be saved.

A week later the same doctor now has 20 patients. She is a little frustrated but they make it work. The team of nurses run around and while exhausted they manage to look after everyone, they lose 2 patients, which is the same 10% death rate as the previous week.

A week later the virus has taken hold in the city, there are now 40 patients. The doctor has never seen anything like this in her life. She struggles as do the nursing team. They lose 7 patients, 3 of which they admit they might have saved if they’d had more time with them.

The following week, there are 80 patients. Additional help has been brought in but it’s difficult, the assistant doctor doesn’t know the hospital or the standard procedures well, the additional nurses are tired from travelling in late the previous night. They lose 17 patients that week.

In week 1 and 2 the death rate in the intensive care unit is 10%. In week 3 it’s 17.5%. In week 4 it’s over 21%.

As long as the virus is spreading those cases will continue to rise. There is a limited supply of doctors, nurses, hospital beds, oxygen tanks. While I’m sure every health professional will always do their best with every patient, it’s basic common sense that given less and less time with each patient, and less resources to help each patient, the percentage dying will keep on increasing.

So many people talk about the Covid-19 fatality rate as if it a fixed number that once properly calculated can determine how we deal with the crisis. I’m certain that nothing could be further from the truth. Each hospital will have a certain number of patients beyond which they simply cannot give each patient the best possible care. Without the best possible care, the chances of dying from Covid-19 increases.

This is all undeniable true – yet you still read people online screaming “tHe dEath rATe IS oNLy 0.1%”. As if the death rate is a feature of the virus. It isn’t at all. As we see with every serious illness out there, the death rate is a combination of the seriousness of the illness and the ability of health professionals to assist the body in fighting it. You only have to look at death rates for various types of cancer to see that they vary dramatically depending on the location and year.

As cases increase beyond a certain predictable point, the death rate will rise. This predictable point can be moved somewhat with an increase in resources (i.e. more doctors, beds and oxygen). However eventually there will be physical limits to what a hospital can contain, and building new hospitals takes times. Cases needing hospitalisation will rise exponentially while any dramatic increase in resources will move on a linear scale. Very few people seem to realise this.

The Fallacy Of Herd Immunity

Throughout the pandemic I’ve read countless times arguments being made that those people of working age (so under the age of 65) should be allowed to go about their normal lives, and eventually contract the virus. Those over the age of 65 (so people typically retired) would hide away and wait for the virus to move through the rest of the population. That would divide the nation between retired older people and younger working people and children. The idea being that as they are statistically unlikely to die from it, younger people can recover swiftly and gain immunity. This then helps the population at large and makes it far harder for the virus to reach vulnerable older people. This is because the virus cannot reproduce in an immune person, and therefore has no opportunity to leap from that person to another.

Again there is some logic to this in principle. In practice it would be a disaster. There are 52.8 million people in the UK aged 64 or under. Let’s ignore the children as they are extremely unlikely to need medical attention. In fact let’s ignore everyone under the age of 30, even though they may still have a nasty illness. So we’re left with 28.8 million people in the UK between the ages of 30 and 65. These are working people who are needed to drive the economy, both in terms of producing goods and indeed consuming them.

(It is not straightforward to get the data, https://coronavirus.data.gov.uk/healthcare?areaType=nation&areaName=England – but here we can see 47,588 of 145,534 patients admitted to hospital with Covid-19 were aged between 18 and 64. I believe we can assume those aged between 18 and 30 were a low percentage of that number. Until I can get better data though I will give a generous 15% of that figure to that age group. In reality it is probably more like 1-5%. So that leaves about 40,000 hospitalisation in England for Covid-19 of people aged between 30 and 65.)

Officially there have been 921,236 cases in England so far. At this point we are admitedly moving to back of the envelope calculations but let’s assume actually 5 million people have had Covid. That leaves around 40 million who definitely haven’t had it. Perhaps 8 times as many as have had it so far.

So that’s 320,000 more hospitalised cases of Covid-19 to come in the next 3-6 months if it is allowed to run rampant. With each person needing doctors to assess them, nurses, drugs, oxygen in plenty of cases. These are people who wouldn’t normally visit a hospital during a typical year remember.

It would very quickly overun hospitals – leading to that ‘low’ death rate quickly getting higher, as people in the 40-65 age group who needed a hospital bed found themselves without.

By the time it was realised that the plan wasn’t going to work, it would be too late. Hospitals being overrun would also have the knock on effect of worsening treatment of everything else in hospital, leading to more deaths there too.

Herd immunity cannot work in a population of millions where even as low as 3-4% lets say of cases will require medical aid. If that number were 0.3-0.4% you would certain consider it, but the facts show that not to be the case.

The solution

As I write several new vaccines are being reported as having had successful trials. This suggests that if enough vulnerable people can be vaccinated then the above calculations will then become much more pleasing to the eye.

To believe we’re at the end of this crisis would be a mistake though. To vaccinate millions of people in a country is not a straightforward task. Even with a huge investment in resouces the most optimistic reports suggest that Spring of 2021 is the earliest that some regions might be able to relax restrictions knowing that a significant number of people will be immune.

Between now and then we have the winter of 2020. One can only hope news of the vaccine doesn’t push people into ignoring the precautions and thus giving new strength to the spread of the virus.

Personally, I can’t imagine I will get offered the virus before next summer at the absolute earliest, and more likely next autumn. Until a very large percentage of a population has been vaccinated I don’t imagine governments will rush to remove movement restrictions and requirements such as a mask.

Of course every day a country is in ‘lockdown’ it is doing untold damage to the economy, both in terms of economic activity and perhaps more worryingly in terms of people’s outlooks for the future. From where I am writing I am a stonesthrow away from two shops that I know have shut down due to a middle aged owner deciding that it simply wasn’t worth keeping the business open. Government aid has paid the rent of many people’s homes for the last 8 months, at the cost of an incredible increase in public debt. This can’t continue without huge tax rises in the future, or significant inflation, both of which will harm economies even more.

I had original planned to discuss the economic implications of the pandemic but I think that deserves a whole blog post to itself.

Thanks for reading all the way through, I know it turned into a long blog post, stay safe, stay extremely vigilant, and look forward to 2022 when I really believe normality will return.

Published inCurrent Affairs

Be First to Comment

Leave a Reply

Your email address will not be published. Required fields are marked *