Coronavirus and the Wisdom of Grandmothers

For the last four years or so, we the Great British Public, have been subjected to a continual lashing storm of hysteria from the poltical, media and commentating classes. And then finally Brexit was all over (bar a few negotiations over trade) and the hysteria subsided into an occasional drizzle. At last the country prepared to draw a collective sigh of relief. And then Coronavirus hit us, shutdown happened, and the whole hysteria thing has cranked up again with scarcely a pause for breath.

There are many arguments and points for discussion. There are lessons to be learned from the origins of the virus and our responses to it. Many people are doing this now, rather than wait for the thing to die down. Comparisons of death rates have been made between the UK and other countries; conclusions have been drawn (often on the strength of incomplete information, or information gathered in wildly different ways from our own data) and armchair experts deployed onto our television screens to decry and denounce the response by the government.

I am not a virologist, or a doctor, or an epidemiologist. But there are a couple of issues that I feel I can make some level of constructive comment upon, by looking at the data that we have publicly available. These two issues of contention are firstly that the disease is out of control in the UK and continues to get worse because the government has been doing all the wrong things. The second issue arises in discussion on social media (but not in the MSM) and is the counter-argument to the fear over daily death tolls. This argument runs something like: “The whole thing is vastly over-hyped; the number of deaths to Coronavirus is a small proportion of the normal deaths and will be swallowed up by the annual totals.”

One of the ways in which the statistics have been regularly presented over the last few weeks has been the cumulative curve of total deaths. Often these have been presented by comparing different countires, such as the one below:

Figure 1: Cumulative curve of total deaths to date, by country:

Screenshot 2020-04-20 at 12.30.30

Diagram taken from: Our World in Data

The problem with a cumulative curve like this is that it is difficult to see when the number of cases begins to decline and we know that we are past the worst. The shape of them means that the disease has only run its course when the curve is actually horizontal. For this reason, a better illustration is to use the daily new cases or deaths, as in the following curve for the UK:

Figure 2: Covid-19 – Daily UK confirmed cases

Screenshot 2020-04-21 at 10.31.00

This graph is the daily reported confirmed cases since 17th march, which is when I started collecting the data. It is correct to data reported on 20th April. From this, we can see that the number of new cases started to flatten out on Day 25, or about the 9th April. In the last three days, it has started to dip, suggesting that the number of new confirmed cases is begining to drop.

However, the graph that everyone likes to look at is the number of deaths due to Covid-19:

Figure 3: Covid-19 – Daily UK reported deaths

Screenshot 2020-04-21 at 11.24.28

This confirms the current thinking that disease has peaked. Because the peak is more obvious, it would seem that the UK peaked (appropriately enough) on Easter Sunday. New deaths ascribed to Covid-19 are declining in number. This seems to be in common with most countries which are reporting figures. We are over the worst.

However, the eagle-eyed amongst readers will have noticed that the vertical scales on these two graphs are not the same. Very obviously, the number of cases reported is much larger than the number of deaths. To put the number of deaths in context with the number of cases, the two curves are plotted using the same scale on the vertical axis:

Figure 4: Covid-19 – Daily UK confirmed cases and deaths on the same scale:

Screenshot 2020-04-21 at 10.55.40

Daily new cases are drawn in red and daily deaths in green. This graph shows the number of deaths is very much flatter (because of the scales) and suggests a much less alarming picture than the deaths presented on their own. This would appear to give credence to those who say that the impact of the disease is nothing like as bad as it has been cracked up to be by the media etc.

To examine this hypothesis, it is necessary to look at the typical pattern of deaths in the UK in normal years, and then to compare deaths in the current year to see if they look normal or abnormal. The following is graph of deaths (for all causes) in England and Wales by week number (where Week 1 is the first week in January).

Figure 5: Covid-19 – Deaths, by Week Number in England and Wales, 2010 to 2019:

Screenshot 2020-04-21 at 12.40.52

This graph shows a curious pattern. For all years, the death rate declines from the drear months of January and February into a patch of uncertain width, caused by such things as ‘flu of varying severity in the seasonal fluctuating warm and cold spells of the early spring.  As the death rate drops still further in the late spring, there is a marked ‘V’-shaped trough which coincides with the Spring Bank Holidays. The curve slopes gently downwards again until the August Bank Holidays, where there is another ‘V’-shaped trough. Then, as the days become shorter and the variable weather of Autumn closes in, the death rate begins to increase again until the week before Christmas. Over Christmas, there is a substantial drop in the number of deaths. After Christmas and when the New Year begins, the death rate massively increases and attains a very high peak in the first two to three weeks in January. From the Christmas trough (which could be as low as 200) to the New Year peak can be as much as 16,000. This is an increase in death rate of up to 80 times in about two or three weeks.

What is extraordinary about this graph (at least to me – I have no doubt medics, insurance companies and undertakers know all about this) is the clear unwillingness of people to die on holiday – and certainly over Christmas. There are some very strong social and psychological impulses going on here which over-ride any physiological shut-down of the human body. It is almost as if many people plan their own deaths. – or at least, when they are NOT going to die.

However, to return to the question “Is Coronavirus causing an exceptional number of deaths in the UK?” The following is the same graph as Figure 5 above, but with the 2020 figures to date, overlaid upon it in red:

Figure 6: Covid-19 – England and Wales Deaths by Week Number, 2010 to 2020 to date (to week ending 10th April 2020)

Screenshot 2020-04-21 at 12.40.17

The year started off conventionally enough, but the moment Coronavirus began to bite, things changed dramatically. It is not necessary to be a statistician or epidemiologist to see that the last two weeks of reported weekly data are extraordinary. The peak is not only much higher than the maxima achieved by even the worst winter mortality rate since 2010. But it is also unseasonal, leaping skywards at a time when the overall trend in normal years is down. This peak will probably begin to decline in the next week of reported figures, but it will remain a big departure from usual for quite a while yet.

The size of this peak confirms the need for extraordinary measures by the NHS to increase ICU beds, along with the associated back-up of ventilators, PPE and so on. There are those who argue that the total deaths this year are unlikely to be particularly high when compared to other years. Looking at the height of that departure, I suspect that this year’s figure will be significantly higher than usual.

It is possible that there will be a lower than normal death rate later this year and early next year, because the most vulnerable people will have already died prematurely during the Cornavirus pandemic. But that remains to be seen. There are, in any case, short-term fears that a ‘second peak’ may occur as lockdown is opened up and vulnerable but currently isolated people are exposed to the virus. There are still many, many unknowns about Covid-19.

To conclude: Firstly, the disease has peaked and the country can be safe in knowlege that the light at the end of the tunnel is growing larger by the day. Secondly, there is no question that this disease has shortened many peoples’ lives and that it has tested the NHS considerably. The NHS has now increased its provision of ICU beds by the construction of the Nightingale hospitals. These have been produced, as if from nothing, in a matter of three weeks or so. It is a measure of the organisational capabilities of our armed forces and manufacturing sector. The NHS and the Civil Service could not have done this on their own.

And finally. Picture the scene in a normal year: It is New Years Day. The family has arrived from all points of the compass and has finished a splendid meal. They are now gathered in the sitting room. The Christmas tree is glowing with lights and tinsel. Sitting in her usual chair in the corner, is Grandmother. Around her are her children, now in their middle age. Various grandchildren, perhaps in their twenties or early thirties, are scattered about with their partners. On the floor are the first tranche of great-granchildren who are  toddling or crawling according to age and inclination. There is a scattering of toys on the floor. The cat has left through the cat-flap in disgust at the noise and chaos. The dog is enjoying himself enormously because at some point, food will drop onto the floor.

Grandmother surveys the assembled family quietly, but with some satisfaction. Were it not for her own strivings – first upon the marital bed, and then upon the maternity bed – none of this lot would be here. You pass her seat, bottle in hand. “I’ll have another sherry if you please,” she says, holding out her empty glass for you to refill. “A large one,” she says sharply, as if you are being a little parsimonious. You dutifully tip a large measure into her glass. “I’ll finish this, and then I’ll go,” she says in a rather final tone of voice.

You are momentarily puzzled. Does she mean “Go to bed early; get an early night?” You frown. “Or does she mean…..? No, surely, it can’t be that – she seems completely healthy.” You study her face briefly. But the sherry glass glass has been swung expertly to her lips and she is now slurping industriously. Her facial expression is unreadable. You pass on, thinking the she has been saying one or two odd things just recently. “Perhaps she doesn’t really know what she is saying”. You comfort youself with that little thought and continue to exchange bandinage with the rest of the family.

But Grandmother knows exactly what she is saying.

And what’s more, the statisticians can prove it.

Statistical Notes:

  1.   Daily new cases and deaths collected by D. Eyles (Figures 2,3 and 4) to 20th April.
  2.   ONS Dataset: Deaths Registered Weekly in England and Wales – Provisional to week ending 10th April.
  3.   Stats plotted using R and ggplot2.
  4.   Figures 2,3 and 4 plotted with smoothed Loess curve.
  5.   Grey band in Figures 2,3 and 4 is the 95% confidence interval for the Loess curve.

2 Comments

  1. I wonder if Figure 4 is compiled using ONS Dataset: Deaths Registered Weekly in England and Wales – Provisional. In which case, if it is the date of registration of a death rather than the actual date of death, then that could contribute to the dips at Christmas and the Bank Holidays when registration offices are closed or people may wait longer than usual to visit the office and register a death.

    1. This issue is a potential problem and may in part explain the ‘v’ shaped troughs. There is a minimum period of 5 days for the death certificate to be signed by a doctor. And doctors on holidays may account for it. But I am not convinced that this explains the whole of the ‘trough’. It needs a bit more work to justify this gut feeling, however.

      Sorry to be slow in responding.

Leave a Reply