Covid-19

aka SARS-CoV-2, 2019-nCoV, Wuhan pneumonia, Coronavirus, The bug

The virus and the disease it causes are known by many names. The World Health Organization (WHO) refers to it as SARS-CoV-2 (Sudden Acute Respiratory Syndrome CoronaVirus 2). In this article I’ll mostly refer to it as ‘the bug’ (or ‘the damned bug’).

• aka SARS-CoV-2, 2019-nCoV, Wuhan pneumonia, Coronavirus, The bug
• Re-write
• TL;DR
• Some notes on the numbers
  • Rates not absolute numbers
  • Public holidays
  • Week numbers
  • Diagnosis
  • Average or normal?
• Just how bad is it?
• Examining the extra death registrations
  • 2020 compared to ‘normal’
  • So what should we have expected in 2020?
• What about the low points?
• The right treatment
• Vaccinations
• From the department of Egregious Statistics and Applied Flannel at the University of West Fantasia

Re-write

The original article had grown out of control so I decided to re-write much of it.

This version replaces the obsolete version which can still be found here.

When I started the original article I expected things to return to normal fairly quickly as people in authority and the general public realised that they had over-reacted. Sadly, I underestimated how popular bad news can be and how reluctant people are to admit they have been misled - and how reluctant leaders can be to admit they have misled the people. As such I didn’t take into consideration how many weeks data I would need to process or how to properly compare data from previous times with current times. I also found I was presenting graphs with too many different styles and colour codes. This re-write attempts to correct these faults and prepare to process published data for at least a further year.

TL;DR

Don’t Panic!

Yes, that’s it. That’s the summary.

OK. Here’s why (for England and Wales unless otherwise stated):

• The high numbers of deaths in April and May 2020 were preceeded by a 2 year period of exceptionally low numbers of deaths. We can assume that many of those who died had survived longer than would normally be expected.
• The lockdown from 23 Mar 2020 seems to have had an immediate negative effect followed by a later, lesser positive effect on death numbers.
• Changes to lockdown rules such as the re-opening of pubs and restaurants and the imposition of mask-wearing have had no effect on death numbers.
• The majority of those who have died with Coronavirus were elderly or had underlying poor health conditions - which is what we see among those who die in normal times. The ratio of ‘elderly’ to other age groups dying during the peak is the same as in normal times.
• The ratio of men:women dying during the peak is the same as in normal times. In normal times men die younger than women.
• From 2020-wk24 (mid June): It’s over. Yes, people are still catching the bug and dying from it - but they’re not catching or dying of ‘flu in the numbers we would expect from this time of year. It doesn’t really matter if you die of Coronavirus instead of ‘flu.
• The numbers quoted on the UK government ‘dashboard’, the numbers quoted by Public Health England (and similar organisations) and the number or death registrations reported by ONS to mention Covid-19 are unreliable. eg In 2020 there were 85,223 death registrations which mentioned Covid-19 on the MCCD (the certifying doctor’s opinion of cause(s) of death) but only 52,814 more deaths than we would expect for the year. This means Covid-19 was mentioned on 31,636 death registrations which we were expecting anyway.

(For those who don’t recognise what TL;DR means - it either means Too Long; Didn’t Read, or Too Long; Don’t Read depending on the context. If you think of yourself as an executive - perhaps with too little time or a generally short attention span - you could take it to mean Executive Summary.)

Some notes on the numbers

Simply counting the number of people who contract or die of any particular condition is far easier said than done and does not allow us to compare changes over time. For example, it might seem that it’s a Good Thing that far fewer people aged over 85 years died in 1841 than in 2018; there were 22 times more deaths of over 85s in 2018 than there were in 1841.

Of course, the reason is that there were fewer people overall and a smaller percentage of them survived to reach their 85th birthday in 1841 when compared with 2018.

Rates not absolute numbers

In order to sensibly compare deaths in 1841 with 2018 we need to look at the rate at which particular subsets of the population died.

Note the almost imperceptible effect of the 20th centuary pandemics (helpfully highlighted grey); they don’t really show up on this scale. We can certainly see a change in infant mortality from 1900 onwards but do note the dramatic change in the graph from 1948… If you’ve ever wondered what a successful public healthcare policy change might look like, this is it; the NHS commenced in 1948.

If we zoom in to more recent times and consider, for example, people aged 85+ years in England and Wales between 1950 and 2018: In 1950 there were 48,477 deaths out of 199,436 people (24.3%) and in 2018 there were 216,781 deaths out of 1,464,421 people (15.0%). But it’s just a little more complicated than even that; there’s a significant difference in death rates between men and women (men tend to die earlier than women).

We can calculate death rates for each population subset for each year and then compare the different years and try to identify trends over time. For example, was 1950 just an exceptionally bad year for over 85s or is there some trend we can identify that leads from the 1950 rate of 24.3% down to the 2018 rate of 15.0%? It probably won’t surprise you to find out that there are trends towards a greater chance of surviving any given year for each age group and sex and a corresponding increase in the average age of death from 1950 onwards. My guess is that these trends are mostly related to free (at the point of use) healthcare, improved medical science, improved public health information (eg anti-smoking campaigns) and improved neo-natal care.

I’m guessing, but I don’t think that improvements in healthcare can keep on increasing our average lifespans at the same rate as has been achieved since the commencement of the NHS. I think I can see a slight levelling off in the slope of the graphs covering the past 20 or so years. We’re now at a time when more than 91% of people in England and Wales were born after the inception of the NHS. If they’ve lived in these countries throughout, then they have had access to NHS healthcare all their lives.

On top of all this if we have the raw data available we can calculate death rates for each age group and sex for each week of each year and identify repeating patterns (seasonal variations) through the course of the average year.

We now zoom in still closer to recent times: ONS has been publishing the required weekly data for England and Wales since January 2010 so we can calculate average rates, identify trends and measure seasonal deviations from these trends using data covering 10 years prior to 2020.

So what are the calculated numbers of expected deaths for 2020 and 2021?

Just to remove any doubt about the above chart and table. It shows (among the all the other figures) that in 2021 we expect on average to find:

• 4.89% of males aged 75-84 die which, if it happens, will be 85,646 men - fathers, grandfathers, great-grandfathers, brothers, husbands, partners, uncles, batchelors, friends, strangers etc.
• 0.01% of males and females aged 1-14 which would be 483 boys and 384 girls - and every one of them a tragedy for someone.
• 14.22% of females over 85 years or 135,147 old ladies.
• On average we expect 552,464 people to die in England and Wales in 2021; a large majority of whom will be over 75 years old.

Armed with knowledge of these trends and seasonal variations we can sensibly compare the number of deaths in each sex and age group against the expected average; eg, we could compare the death rates of men aged 85+ in week 16 of 2020 with the average death rates for men in the same age group in the same week in years 2010-2019 (though we need to be cautious about 2020-wk16 because it includes a public holiday) - and so on for each week and subset of people.

Public holidays

But what are those strange dips in the seasonal graph? Public holidays present complications and delays for death registrations. Register offices in the UK typically observe weekends and public holidays which can make it difficult to get an appointment to register a death at certain times of the year. Most notably around Christmas, New Year and Easter.

• The date of Easter is highly variable and calculated based on phases of the moon in relation to the calendar. It’s defined that:
  
  Easter falls on the first Sunday after the Full Moon date, based on mathematical calculations, that falls on or after March 21. If the Full Moon is on a Sunday, Easter is celebrated on the following Sunday.
  
  
  • It follows that the week numbers for the two Easter public holidays (Good Friday and Easter Monday) are also highly variable. The earliest possible date for Easter Sunday is 22 Mar and the latest is 25 Apr.
• In the UK the secular public holidays of ‘Early May Bank Holiday’ (week 18 or 19), ‘Spring Bank Holiday’ (week 22 or 23) and ‘Summer Bank Holiday’ (week 35 or 36) are usually scheduled to fall on a Monday. For 2020 the Early May Bank Holiday was deferred to the following Friday to coincide with VE Day (8 May).
• In the UK if the fixed-date public holidays of Christmas, Boxing Day and New Year fall on non-working days (ie Saturday or Sunday) the public holiday is deferred until the next normal working day (eg 26 Dec 2020, Boxing Day fell on a Saturday so the public holiday was deferred until the following Monday, 28 Dec 2020).

The effect of register office holidays is most obvious in the delays of death registrations around the beginning or end of the year and the mid-year secular holidays. Although the Easter holidays do disrupt register offices the variability in the date blurs the effect when data from multiple years are averaged. We need to remain cautious when comparing data relating to weeks near these public holidays as they can introduce delays in registration (but obviously not significant changes in overall numbers of deaths per day). To deal with the Christmas/New Year holidays by far the most reasonable approach is to average the death rates for weeks 51, 52, 53 (if there is one), 1 and 2 and compare the weekly rate with other years. The Coronavirus Act 2020, Schedule 13 (effective for 2 years from 25 Mar 2020, 2020-wk13) has (perhaps) simplified the process of the registration of deaths so that public holidays have less impact on the numbers than previously.

¹ New Year public holidays for 2010, 2016 and 2021 were actually included in the final week of the previous year
² Extra public holiday in week 17 of 2011 to celebrate the wedding of Prince William and Catherine Middleton
³ Extra public holiday in week 23 of 2012 to celebrate the Diamond Jubilee of Queen Elizabeth II
⁴ Christmas and Boxing Day public holidays usually occur in the same week but did not in 2015 and 2020
⁵ In 2020 the Early May public holiday was moved to Fri 8 May in order to celebrate the 75th anniversary of VE Day - celebrations were muted by lockdown restrictions

Week numbers

By international convention (ISO 8601), Week 1 is the first week (defined as Monday to Sunday) of the year in which the Thursday (the middle day of the week) falls in January. This means that Week 1 can start as early as 29 Dec or as late as 4 Jan; and that the last week of the year can end as early as 28 Dec or as late as 3 Jan. This in turn means that the New Year day holiday is sometimes counted in the last week of the previous year rather than in the first week of the current year. It also means that although we usually have 52, we sometimes get 53 weeks in the year - as we did in 2009, 2015 and 2020.

Diagnosis

A complication in only counting cases or deaths due to a particular disease is consistent diagnosis. If a disease or cause of death has no agreed definition then doctors cannot diagnose it consistently. As an example of this, AIDS (the disease caused by HIV infection) was not recognised until 1982 so no doctor could put it on a death certificate prior to that. That doesn’t mean that nobody died of it in 1950 - just that it was never recorded as such. The same is true these days: if someone has difficulty breathing and then dies, the certifying doctor is going to suspect COVID-19 because that is in the forefront of their mind. No need for a test. Put it down as ‘probable COVID-19’. Related to this the WHO updated their ‘COVID-19: Case Definitions’ on 16 Dec 2020 - presumably if you didn’t match the previous definition then you didn’t actually have COVID-19 - but now you would.

There have been further changes which make comparing death diagnosis data from other years with 2020 problematic. To register a death we still need to provide an MCCD (Medical Certificate of Cause of Death) signed by a medical doctor. However, there have been a few changes in 2020:

• A medical certificate can be accepted from any medical practitioner so long as they are able to state to the best of their knowledge the cause of death. Previously the signer had to have ‘attended’ the deceased in their last illness.
  • A doctor is now considered to have ‘attended’ the deceased if they have carried out a physical or video consultation recently. Previously they had to have attended in person.
  • Any doctor can sign the certificate provided there is an ‘acceptable’ cause of death and they know that some other doctor ‘attended’ the deceased recently.
• A doctor can sign the MCCD even if they have only attended the deceased after death - but not by video in this case.
• Appointments to register deaths are now usually completed by phone with the MCCD submitted electronically and other paperwork to follow. This helps reduce the delay between the date of death and the registration date.

As definitions change from time to time it means that we can not reasonably compare diagnosis data gathered before with data gathered after the change. Another example would be attempting to compare the number of Covid positive test results before and after the introduction of mass testing. It follows that the only reasonable measure of the progress of any disease is the impact on the all-cause death rate while recognising that:

• It does not measure how much effort (and success) is going into preventing deaths attributed to the bug.
• Death is the most extreme endpoint, the vast majority of people who catch the bug will not die because of it.
• Because death is an endpoint, it does not measure how many are currently suffering with or have recovered from the bug.
• It gives no measure of how many suffer long-term health damage as a result of the infection.
• Using registered deaths as our measure introduces further delays. Deaths are usually registered within 5 days of the event, but under some circumstances (including referral to a coroner’s court or a lack of available register office appointments near public holidays) registration can be delayed beyond that.
• ONS usually produce their reports of registered deaths 11 days after the Friday of each week.

Average or normal?

Another complication in comparing numbers over time is the crucial difference between ‘average’ and ‘normal’. Our example here is considering deaths of 1-14 year olds:

• In England and Wales the average number of 1-14 year olds who died each week during 2010-2019 was 19.66. That’s nearly 20 kids every week for 10 years.
  • If you were told that in the week ending 28 May 2010 (2010-wk21) there were 32% more deaths than the long term average (2010-2019) for this age group would you be more shocked? Should there be a public inquiry?
  • If you were then told that in 79 weeks of that 10 year period deaths in this age group were 25% or more above the average would that surprise you? (15% of the time deaths were more than 25% above average).

The issue here is that the death of a 1-14 year old is really a very rare event. On average fewer than 20 of these children die each week in England and Wales. A single event such as a car crash or other accident or even a crime can cause a significant spike in any given week. While the average is 19.66 children’s deaths per week it is quite common for there to be 29 or more or 10 or fewer. This variablity in the data can be measured and assigned a ‘score’ called the Standard Deviation (SD) of the data. The higher the SD score, the more scattered/variable the data is. The SD of the numbers of deaths of 1-14 year olds in England and Wales in 2010-2019 is 4.99 which is very high compared to the average of 19.66. This can be taken to mean that the average of 19.66 plus or minus two SDs (ie 10-29 deaths per week; or the average +/- 50%) should be considered entirely normal.

Generally, in mortality statistics if any single result is 4 standard deviations above or below the average then we can assume something rather unusual has happened. EuroMOMO takes this approach when comparing the death data from many European countries including the four countries of the UK. With this in mind we should recognise that:

• At no time throughout 2010-2020 did the numbers of children’s deaths exceed 3.7 SD above average.
• The greatest number (38 children’s deaths, 93% more than average, +3.7SD) were registered in the week ending 10 Aug 2012 (ie nothing to do with COVID-19).

When comparing numbers, we need to calculate Pecentages (rates), Averages, Trends and Standard Deviations. Without these tools we can’t hope to understand what is going on. Don’t worry though, we don’t have to do the thousands of arithmetic calculations directly - it’s far easier and less prone to error to use a spreadsheet to do the repetitive work.

Just how bad is it?

It was bad but we’ve had worse in living memory.

*Notes:

• The majority of the data for this chart is drawn from Office of National Statistics (ONS) Excess Winter Mortality datasets for England and Wales combined with their population estimates and population projections (blue columns).
• ONS’ definition of Excess Winter Mortality seems odd to me. They measure the deaths in the Winter months in comparison to the preceding Summer/Autumn and following Spring/Summer. This means that there is no concept of ‘normal’ extra Winter deaths in their numbers so we have to unpick them to find what I would call the true ‘excess’.
  • By applying the ONS definition to a 10-year average (2010-2019), it is usual to have about 21,000 ‘excess’ deaths (ONS’ definition) in a Winter season. This can be a useful comparison to see if the ONS reported excess is higher or lower than usual.
  • Excess Winter Mortality is a purely retrospective measure; it cannot be calculated until the following August. It is far more useful to be able to predict an expected number of deaths per week and compare weekly progress against that.
• I predicted that ONS would not use their usual calculation when they reported the Excess Winter Deaths for the 2019-2020 season. They excluded any deaths where Coronavirus was mentioned on the death certificate when calculating the non-Winter baseline.
• The calculations I have used to estimate recent excess deaths (red/yellow column only) are mine - don’t blame ONS if I have made invalid assumptions in the calculation. I have calculated that in weeks 11 to 28 (about 4 months) there were 60,083 more deaths than the 2014-2018 five year average.
• Data for the COVID-19 (red) column is drawn from the ONS weekly dataset count of deaths registered in the four months since week 11 (beginning 7 Mar) of 2020 to week 28 (ending 10 Jul) in England and Wales which mention COVID-19 on the death certificate.
• The ‘lockdown’ (yellow) column is the difference between the cumulative excess deaths since week 11 and the COVID-19 column. As such it represents excess deaths which are not apparently explained by the Covid bug.
• As at 12 Aug 2020 UK government agencies reckon that 41,329 people have died from COVID-19 in the whole of the UK - revised down from 46,706 due to a recent change in the way Public Health England (PHE) counts COVID-19 deaths.

There have been 50,505 deaths registered in England and Wales where COVID-19 was mentioned on the death certificate in the four months since such deaths were first recorded. This just exceeds the ONS reported figure of 49,410 excess Winter deaths in Winter 2017/2018. The figures suggest 2017/2018 was quite a bad season (highest absolute number of excess Winter deaths since 1975/1976) but I certainly don’t recall any media panic - at least not until after it was over. When we express the reported excess deaths as a percentage of the population at the time, COVID-19 is the 31st worst 4-month peak in the last 70 years in England and Wales. To avoid any misunderstanding note that I am comparing a 4-month peak above the 5 year average for the same time of year (COVID-19, mid-March to mid-July) against 4-month peaks above the rest of the corresponding year (excess Winter deaths, December-March above August-November and April-July). Excess Winter deaths are normal and expected - COVID-19 was unusual.

Examining the extra death registrations

The ONS EWM figures suggest something unusual happened in the Winters of 2014/15 (43,720 ‘excess’ deaths by ONS’ reckoning) and 2017/18 (49,410 ‘excess’ deaths). Indeed, if we calculate and chart the difference between the actual and expected numbers of deaths for each sex and age group for each week from week 1 2010 (taking into account trends and population changes) we can see clear clusters of unusually high numbers of death registrations around Winter 2014/15 (~14,000 extra deaths by my reckoning) and 2017/18 (~16,000 extra deaths) and unusually low numbers at other times. Of course, by far the most striking feature of the 11-year chart is the large spike in late March to early June 2020 - but we’ll come back to that in a moment.

A few important things to note or bear in mind when considering these charts:

• These are charts and graphs of differences in actual number of death registrations from an expected, calculated number. So a positive spike represents more deaths than expected and a negative spike shows fewer deaths than expected.
• These are charts and graphs of death registrations. The actual date of a death will typically be about a week previous. If the death is caused by an infection such as ‘flu or the COVID-19 bug then the date of infection will be even earlier - by another couple of weeks or so.
• Although the above chart takes normal seasonal variations into account, the majority of significant spikes (positive and negative) occur around the Winter months.

Compare the shape of the positive spikes in the Winters of 2014/15 and 2017/18; the 2014/15 spike has a distinctive shape.

Note the sharp rise and gradual tailing off of numbers of deaths in Winter 2014/15; also the peaks are taller and narrower than for 2017/18. If we plot the cumulative extra deaths for each age group from 2014-wk50 to 2015-wk18 we see a remarkable pattern - the graph is a good fit for ‘Gompertz’ curves (the fine dotted/dashed lines) for the top three age groups.

However, when we look at Winter 2017/18 in a similar way there is no correspondence to similar Gompertz curves.

So, what’s the significance of fitting Gompertz curves? Gompertz curves are sometimes called ‘growth curves’ and are found in many natural processes such as the growth of tumours and bacterial colonies and in the course of epidemics.

• We can assert that Winter 2014-15 extra deaths were mostly driven by a single epidemic. In their 2014/15 EWM report ONS blame the ineffectiveness of the ‘flu jab that season.
  • The top three age groups correspond to Gompertz curves with the same timing for the peak death rate (steepest point).
  • Peak increase in registered deaths was around the week ending 16 Jan 2015.
  • Actual deaths will be about a week before registration.
  • Infection will have occurred about two weeks prior to death (on average).
• Winter 2017-18 extra deaths were driven by multiple factors; the death registrations do not follow an identifiable epidemic curve for any group.
  • Causes of extra deaths were probably driven by the severe Winter weather (‘The Beast from the East’ and ‘The mini-Beast from the East’ from 22 Feb to late-Mar 2018).

So what about that giant spike in 2020-wk13 to 2020-wk23? What does that one look like?

A damn-near perfect match to Gompertz curves.

• Yes, that’s an epidemic with peak increase at around week 3-4.
  • It’s a nasty one too - over 53,000 extra people dead in a period of just 12 weeks. No wonder people got scared.

Now let’s look at the tail end of 2020. 16 Oct 2020 (2020-wk42) onwards.

No, that’s not an epidemic; it looks like multiple competing factors. I don’t care what these people were diagnosed with; these are not predominantly single-epidemic deaths.

It’s important to note that the formula for a Gompertz curve includes two opposing exponents. One drives up the number of deaths (in our examples) in the early part of the curve and the second drives down the number of deaths later in the epidemic. The general formula for a Gompertz curve is:

a * EXP(-b * EXP(-c * time))

The parameters b and c do not represent anything particular in the real world but a corresponds to the total deaths reached once the curve levels off. We don’t observe values for a, b or c directly but we derive them by finding the best ‘fit’ for the observed data. If the data doesn’t ‘fit’ a Gompertz curve then the data does not represent a natural process with the same conditions throughout.

2020 compared to ‘normal’

As mentioned above calculating the Standard Deviation of a set of numbers can give an idea of how widely scattered they are and how far a measurement has to be from the average to be considered ‘abnormal’. In mortality statistics it’s the convention that if a measurement is 4 SDs above or below average then something significant has happened.

If we look at the Spring 2020 epidemic peak in these terms we see this:

• Male age groups over 45 years and female age groups over 15 years show a peak above 4 SD in week ending 24 Apr 2020 (2020-wk17); clearly a very significant event.
• Older age groups have higher peaks.
• The Easter holidays (Fri 10 Apr 2020 and Mon 13 Apr 2020, 2020-wk15 and 2020-wk16) appear as a less steep rise in the peak.
• The dip between the two peaks coincides with the Early May Bank Holiday (Fri 8 May 2020, 2020-wk19).

If we look at the last few months of 2020 and into 2021 in a similar way we see this:

• The gap in the graph is at 2020-wk53. Because there was only one ‘Week 53’ in our baseline period of 2010-2019 (week ending 1 Jan 2016, 2015-wk53) there can be no valid Standard Deviation with which to compare that week’s numbers (the SD of a single sample is zero).
• The graph is chaotic; there is no well defined peak common to the different sex and age groups.
• The oldest age groups (males and females over 85 years) do not show large peaks and are certainly not the highest peaks.
• The greatest increase seems to be among age groups 45-64 and 65-74.
  • The increase for males 45-64 is even greater than in the Spring peak: 2020-wk17=+11.38 SD, 2021-wk3=+12.93 SD
  • The most significant increase is among later-working-age males.

For the analysis it really does not matter what these deaths are diagnosed as; whatever these people are dying of it’s not the same as the Spring peak. However, from a harm reduction point of view it’s important to try to understand the cause - and whatever it is it’s not the epidemic.

So what should we have expected in 2020?

Finally (for now) let’s look at the reported numbers compared with the calculated/expected numbers for England and Wales in 2020:

• In the massive peak in the eleven weeks from 2020-wk13 to 2020-wk23 (late March to early June):
  • We expected 114,905 registrations but there were 168,396 deaths registered; 53,491 (46.6%) more deaths registered than expected.
  • Doctors mentioned COVID-19 on 46,996 death registrations during this peak (27.9% of the peak total).
    • Either there were 6,495 extra death registrations with non-COVID-19 causes or these were misdiagnosed COVID-19 related deaths (or a mix).
  • UK Government dashboard reports 35,472 COVID-19 deaths.
• Death registrations were also above expectations from 2020-wk42 (mid October) onwards.
• From 2020-wk43 (week ending 23 Oct 2020) ONS started to include the number of instances where COVID-19 is stated on the death certificate to be the main (they use the term ‘underlying’) cause of death (as distinct from a contributary cause).
• In the eleven weeks from 2020-wk43 to 2020-wk53 (the end of the year):
  • We expected 115,252 registrations but there were 129,878 deaths registered; 14,626 (12.7%) more deaths registered than expected.
  • Doctors mentioned COVID-19 on 27,130 death certificates (20.9% of the period total) but stated that it was the main cause of death on 23,582 certificates (18.2% of the period total).
  • There were 14,626 more death registrations than expected but 23,582 where the main cause was COVID-19.
    • Excluding COVID-19 as a main cause of death there were apparently 8,956 (-7.8%) fewer deaths in this period than expected.
  • Where COVID-19 was mentioned on the death certificate it was listed as the main cause 87.2% of the time.
  • UK Government dashboard reports 29,466 COVID-19 deaths.
• Across the whole of 2020 in England and Wales:
  • We expected 561,300 registrations but there were 614,114 deaths registered; 52,814 (9.5%) more deaths registered than expected.
    • 52,814 extra deaths corresponds closely to the number of extra deaths registered in the peak in weeks 13-23 (53,491 extra death registrations).
  • Doctors mentioned COVID-19 on 85,223 death registrations.
  • UK Government dashboard reports 70,083 COVID-19 deaths
• The mismatch between the UK Government dashboard, the cause of death put on death certificates and above/below expectation numbers of deaths shows that the diagnosis of COVID-19 as a cause of death is very unreliable.

What about the low points?

If we look again at parts of that 11-year chart we see that directly before the Spring 2020 ‘spike’ we have a long period of lower than expected deaths. Over the period 2018-wk17 to 2020-wk12 (100 weeks, almost 2 whole years) there were nearly 40,000 fewer deaths than we would expect. Think about it: 40,000 people lived rather longer than we would expect - and then came COVID-19.

So what preceded the Winter 2014/15 epidemic? A period of unusually low deaths from 2013-wk24 during which about 21,000 fewer people than expected died.

This shouldn’t surprise anyone. It’s inevitable with an average-based analysis like this; ‘lows’ must be followed by ‘highs’. However, do bear in mind that the trend/average has been drawn from data over the 10 years from Jan 2010 to end-Dec 2019 (ie it specifically excludes 2020) - and still it neatly fits in with the extreme event of the Covid epidemic.

The right treatment

This is an allegorical section - not based on hard facts such as the number of death registrations.

• Medicines are strictly controlled in most developed countries.
  • The authority to prescribe most medicines is reserved to qualified and registered medical practicioners.
  • A medical practioner without proper registration can get into serious trouble.
  • It is illegal in many countries to claim that a product provides medical benefits without providing adequate evidence. If you successfully prove your product is a ‘medicine’ then it must adhere to medical quality standards and is subject to monitoring in the population.
  • As with all such things, there are a few exceptions which can be self-prescribed (eg aspirin, paracetemol, ibuprofen, cetirizine, loretidine). These are often subject to strict controls over the amount you can buy over the counter at any one time (eg maximum 2 packs of 16 x 500mg paracetemol in the UK).
• Inappropriate prescribing of medicines can have serious consequences:
  • Side effects that are worse than the condition being treated.
  • Side effects that require some other medication to alleviate them.
  • Addiction risks.
  • Diminishing returns - requiring more and more of a drug to acheive the original benefit.
  • Risk of self-harm/suicide.
  • Building up disease drug resistance (eg MDR-TB (Multi-Drug Resistant Tuberculosis) and drug resistant malaria).
  • Diagnosis/drug mismatch (eg prescribing blood thinners to an haemorrhagic stroke patient).

I’ve demonstrated above that the NHS is a Good Thing. Partly to try to protect its budget from abuse we have ‘NICE’, the ‘National Institute for Health and Care Excellence’. (No, I’ve no idea why the ‘H’ for Health does not appear in their acronym. I also don’t know which Nation of the UK they’re referring to). NICE attempt to further control what medicines NHS practicioners can prescribe based on a complicated cost/benefit analysis.

Before a medicine can be approved for use in the NHS it has to be proved to be:

• Safe for use as recommended - or at least with managable harms which are ‘worth it’ for the patient.
• Effective - or at least better than placebo
• Pass a benefit/cost assessment. It may be that a slightly less effective treatment that costs considerably less or has lesser harms will be approved rather than a new drug.

Once a medicine is approved its use and performance are carefully monitored. Approval can be withdrawn if later analysis contradicts any of the three tests above.

The potential for harm if we get this wrong is enormous. In 1957 a drug which was an effective anti-nausea treatment for motion sickness was then taken by many to control morning-sickness - with disastrous consequences. We must never forget Thalidomide. More recently in 2009, the Pandemrix vaccine which was developed as a preventative for the H1N1 ‘swine flu’ influenza pandemic has been associated with very severe adverse reactions in a small minority of subjects.

So, where am I going with this? Of course, you’ve probably seen this coming a mile off: Do our various governments’ responses to the Covid epidemic pass the three tests?

• Safe for use? From a ‘medical health of the nation’ pespective I strongly doubt it but can’t prove it (yet). For England and Wales we’ll have to wait at least for the annual deaths report for 2020 from ONS - probably around August 2021.
• Effective? No. I see no evidence in the death rates that lockdowns and near lockdowns and most other interventions have been effective. I’ve shown above that looking at anything other than all-cause death rate is unreliable. We need to see a (beneficial) change in the graph which has a plausible link to the time of an intervention (like the step down in the graph of neonatal deaths at the inception of the NHS).
• Benefit/Cost assessment? No. I see no evidence they’ve even gathered data to try to make such an assessment.

On these results these lockdown and social distancing ‘treatments’ should be withdrawn from use. They’ve got as much credibility as homeopathy. If a treatment is not effective why keep prescribing it? Perhaps they think it’s a placebo?

Vaccinations

I am generally a fan of vaccines. The introduction of vaccines against smallpox, polio, measles etc seems to be associated with lower death rates among children and adults and longer lives on average (ie later death). However, I’m not convinced that the offer of annual ‘flu vaccination for all over 65’s in the UK (introduced in year 2000) has had any measurable benefit (if anything the gradual improvements in health over the past 70-odd years have levelled off a bit since 2000). In my mind this lack of benefit is because ‘flu vaccination is aimed squarely at the elderly and so cannot make the significant impact that childhood vaccination does.

Vaccines are rather a special class of medicine. They are intended to be administered to healthy people as a preventative. They are not typically given to someone already suffering from the relevant infection. Childhood vaccinations given in the UK typically give lifelong (or long term) immunity from very severe diseases (around the world polio and measles still result in many becoming disabled for life and before it was eradicated smallpox was a notorious killer). As with any medication, vaccination is not without risk but the damage to the kids caused by very rare severe side effects is on average much less than the uncontrolled diseases would cause.

So bearing in mind that there is a small risk is it ethically OK to vaccinate children against ‘flu not for their own benefit, but to try to prevent the spread of ‘flu to protect other (elderly) people? If that is acceptable to you, is it also OK for the State to insist on it? If that is acceptable too, what should the penalties be for parents who refuse the vaccination for their kids (keep in mind it’s not for the kids’ benefit)? How many children suffering vaccine damage is tolerable to protect Granny? We’re heading into an ethical minefield here.

Kids are not able to make their own decisions on this sort of thing; medics must get approval (informed consent) from parents or guardians before they proceed with vaccination.

Why are we considering giving kids Coronvirus vaccinations? It’s not for their own good; I’ve shown above that on average kids are not badly affected (in 2020 in England and Wales there were 7% fewer deaths than would be expected among 1-14 year olds). It seems we’re prepared to treat our kids so that we can visit other countries for a holiday. I have no reason to believe the risk from the Coronavirus vaccine is any higher or lower than from the ‘flu vaccine - but there’s no good reason to expose kids to that small risk.

However, there is one more very good reason not to give the vaccine to kids and those not statistically vulnerable to Covid-19: In the UK we are told by our Government that even if we are vaccinated we can catch the bug and pass it on. In other words it is only useful for preventing serious illness in the recipient - not for control of the spread to vulnerable people. So, there’s no point in jabbing the kids.

From the department of Egregious Statistics and Applied Flannel at the University of West Fantasia

The BBC published this article in which among other things they report research that shows that a person aged 85+ is 630 times more likely to die of Covid than a person aged 18-29. The arithmetic is probably correct but it’s almost completely meaningless. From the (pre-Covid) 2010-19 annual death statistics for England and Wales I have calculated that if you were a woman aged 90+ your risk of dying (of anything) was 1,346 (one thousand three hundred and forty six) times higher than that of a woman aged 20-24.

Unsurprising fact: as we get older we are more likely to die.

Here’s an idiotic thought: if your risk of dying from Covid is lower than your general risk in life would it lower your risk of death to catch it? Please don’t try this at home.

Feel free to copy/paste as much of this text and/or graphics as you would like for any reason at all. If you want to link to it I would advise making sure that the Wayback Machine (Internet Archive) has a copy so that if I change my text/graphics you can refer to what this post used to say. If you want to tear my analysis apart please do so. If you do refer to my post in any way I’d be pleased if you would let me know so that I can learn from you.