Covid-19 excess deaths, ‘dry tinder’, and the winter deaths seesaw

How many excess deaths has Covid-19 caused in England and Wales? This is a trickier question than you might think. It isn’t just the number of people who are listed as a Covid death on the UK government’s website, which as of this day or writing (18th August) is 38,331 (it’s 41,381 for the whole UK.)  As we know, many of these people died with Covid, but not of Covid. (In fact, judging by the numerous stories about doctors determined to put Covid on the death certificate without any good evidence, it seems that some of those people didn’t even have Covid.) We have no hard figures on how many people were really killed by Covid (and in fact it’s a difficult matter to decide what counts as ‘being killed by Covid’), but it’s unlikely to be 38,331.

But this isn’t the issue that I am concerned with in this article. I am concerned with excess deaths, that is, deaths that are over and above what you would normally get in a period without the disease or condition that is under examination. How many people have died from Covid who wouldn’t have soon died anyway? We know that Covid kills a lot of medically vulnerable people who were probably going to die soon anyway, or who were going to be finished off by the flu, etc. How many people were killed by Covid who weren’t in that condition? They’re the excess deaths. That’s the number we should be concerned with, not the number of Covid deaths.

ONS excess deaths

You might think this problem can be solved by looking at all-cause mortality excess deaths. The Office for National Statistics records, on the weekly deaths spreadsheet it puts out, the number of people who died each week, and also the five-year average for that week. Using that it can be worked out that since the start of the Covid crisis in week 11 there have been 57,635 excess deaths. But that figure is clearly not what we’re after either, because it will include all the extra deaths caused by lockdown itself, all the heart attacks caused by people too frightened to go into hospital, the deaths caused by cancer patients who had their treatment cancelled, and so on (which is why the excess deaths number is greater than the Covid-19 deaths number).

But, it might be said, the number of all-cause excess deaths, being higher than the Covid deaths number, tells us that the excess Covid-19 deaths number is just the same as the number of Covid deaths (while the difference between the overall excess deaths and the Covid deaths is the number of ‘lockdown’ deaths).

But this is too simple. There are two (related) problems with it. The first is that we don’t normally get diseases of this sort in spring. Normally they occur in winter. If this disease had happened to occur a few months earlier then it could have had the same effect as it did in April and May, but our methodology would not have produced anywhere near so many excess deaths in that scenario, because there are always more deaths in the winter months than spring, as this is the period when flus and similar diseases usually take hold, and this means that the five-year average for weekly deaths is higher in winter. In fact, the difference in the five-year average between winter and spring is often 3000-4000 a week. Thus the excess deaths calculated by this method could have been far less had Covid appeared in winter than spring, if we had had a winter without many flu deaths. (For a great discussion of how things would have looked had Covid occurred in winter rather than spring see Alistair Haimes writing on this site in May.)

It makes no sense to use a measure of excess deaths that could give very different answers for two possible scenarios, even though in both scenarios the exact same people, in the exact same physical condition, were killed in the exact same way. So I conclude that using the five-year average in this way isn’t a good method of getting a proper handle on excess Covid deaths, at least in this situation.

Excess winter deaths

The other issue that needs to be taken in to account here is the cyclic nature of excess winter deaths. Excess winter mortality, or EWM, is the phenomenon referred to above: more people die in the December-March period in the UK than the rest of the year. This phenomenon is  observed in many other countries, not just the UK.

Helpfully the ONS publishes yearly records of the EWM for England and Wales going back to 1950. (I thank Hector Drummond Magazine reader Laurence Hodge who first let me know of this.) I have graphed this data for you. (The actual numbers are available in Appendix A. Note that all graphs can be clicked to enlarge.)

 

 

The orange line is the moving five-year average for EWD, which is calculated from two years before to two years after, which is why it starts two years after, and finishes two years before, the blue line.

Scientists aren’t entirely sure what causes the excess of deaths in winter, but generally it is thought that the main reason is influenza and other viruses which thrive in winter, and which finish off a lot of old people who are vulnerable. (Flu mainly kills older people, although it kills more young people than Covid does.) Even the common cold can finish off a lot of people who are near death.

(You might think cold temperatures play a part too, but actually the link between cold temperatures and EWD is not at all straightforward. Lower vitamin D levels caused by less sunlight may also be involved too. But these are issues for another time.)

The EWD seesaw

One thing we can notice right away from the graph is that the blue line, the number of deaths, zig-zags hugely from year to year around the average, often going up to the 40,000 mark and sometimes 50,000 and beyond, and often coming back down to the low 20,000s the year after. Occasionally you get two high years, or two low years, but mostly it it’s a yearly see-saw.

Some have speculated that probably a major factor in this see-saw pattern is that in a winter where the flu fails, for whatever reason, to get much of a foothold, or the dominant flu strain going around is a milder one, then many elderly people who would have been killed in a bad flu season survive it, and make it to next winter (especially if there is no heatwave over summer, which can also cause excess deaths). Such people are those who are physically very frail, or who have serious medical conditions, or who have weak immune systems (immune system efficacy normally declines with age), and who are generally very vulnerable to respiratory diseases. Luckily they make it through that less-bad winter, but the next winter they are killed by a worse strain of flu, or by a flu that managed to spread more that winter.

I have seen this explanation referred to on social media as the ‘dry tinder’ theory. It’s not a very nice term, but it does express the essence of the theory well.

Whether or not this really is the explanation of the zig-zag pattern (or, more likely, a part of the explanation), it is clear that a winter with low excess deaths is likely to be followed by a winter with high excess deaths. And if not that winter, then it’s likely that the next winter again will have a high number of excess deaths. There are a few exceptions to this (for example four low winters from 2001-2003-04), but in general it is true that a low winter is likely to be followed by a high winter, and two low winters is even more likely to be followed by a high winter.

We can see from our graph that winter 2018-19 was a low winter. In fact, at 23,200 it had the fourth-lowest excess winter death toll in the last seventy years. So most likely we were due for a large rebound in winter 2019-20. Could some of the Covid deaths that we have seen in spring 2020 be deaths that were likely to have happened anyway had we been hit with a bad flu in winter 2019-20? Was the winter just gone a mild winter for deaths? If so, if we had two mild winters in a row (‘mild’ here refers to death numbers, not weather), then wouldn’t we have had an awful lot of ‘dry tinder’ for Covid to ‘burn’ through in spring?

Getting clearer on what ‘excess winter deaths’ is

Before we can answer that question, though, there’s something we need to get clearer on. What the ONS calls ‘excess winter deaths’ are not what you may think they are. They are not the actual numbers of people who died that winter over and above an average (such as a five year average) for that time of year. In actual fact these numbers are relative numbers. More specifically they are the numbers left over when you subtract the non-winter deaths from the winter deaths. The winter deaths are taken from the start of December to the end March, and the non-winter deaths are the preceding August to November and the following April to July. The ONS says:

our standard method defines the winter period as December to March and compares the number of deaths that occurred in this winter period with the average number of deaths occurring in the preceding August to November and the following April to July. The calculation used is:

EWD = winter deaths – average non-winter deaths [section 11]

As there are, in the ONS’s categorisation, eight non-winter months and four winter months, we can presume that by ‘average’ non-winter deaths they mean that they halve the numbers for the non-winter months (it is typical of the ONS to be vague about this sort of thing).

So these are relative numbers that depend partly upon the number of winter deaths, but also partly upon non-winter deaths too, and the latter can vary from year to year, not as much as winter deaths do, but to some extent. Look at this graph that David Clark did for another blog post of mine to see this.

 

 

Raw winter death numbers

To get a more accurate handle on winter deaths I realised I would have to look at the raw data for the December to March period. The ONS publish monthly death totals going back to 2006, so I added together the December to March totals for each winter . I then added in the EWD totals. Here are the totals.

Winter	Deaths	EWD
2019-20	197,551	Not yet available
2018-19	185,191	23200
2017-18	209,689	49410
2016-17	199,353	34410
2015-16	187,652	24500
2014-15	205,277	43720
2013-14	174,232	17280
2012-13	184,136	31150
2011-12	179,138	24040
2010-11	181,845	25960
2009-10	180,318	25660
2008-09	192,467	36330
2007-08	180,332	24620
2006-07	179,749	23540

 

And this is what these numbers look like when graphed.

 

 

It can be seen that the EWD numbers roughly track the raw numbers, only at a lower level. To make this clearer, I worked out the differences from one winter to the next for both the raw numbers and the EWD numbers, and this was the result.

Winter	Difference-raw	Difference-EWD
2018-19	-24,498	-26,210
2017-18	10,336	15,000
2016-17	11,701	9910
2015-16	-17,625	-19220
2014-15	31,045	26,440
2013-14	-9904	-13,870
2012-13	4498	7110
2011-12	-2707	-1920
2010-11	1527	300
2009-10	-12,149	-10670
2008-09	12,135	11,710
2007-08	583	1080

 

I then graphed these numbers.

 

 

While the numbers aren’t exactly the same, they’re close enough to enable us to presume that the EWD numbers are a pretty good proxy for the absolute winter numbers. (In fact, they’re much closer than I expected them to be). But anyhow I shall mainly make use of the absolute numbers in the rest of my analysis.

What should we have expected in winter 2019-20?

So: back to the ‘dry tinder’ idea, the idea that some of the 2020 Covid deaths were deaths we would have expected anyway from the usual yearly oscillation, especially if the winter of 2019-20 had not seen a high number of deaths.

First of all, we can ask how many excess winter deaths could we have ordinarily expected in winter 2019-20 (assuming a world where there was no Covid)? Here, though, we run into a problem. There’s no way of getting anything more than a large, ballpark numerical range for this. We know that there’s a fairly regular rebound pattern, but the numbers vary greatly, in a semi-random fashion. There’s no great regularity to how bad the bad winters are, so really, it’s impossible to be precise. But we could say, with some confidence, that most likely 2019-20 would be a bad winter, and it’s more probable than not that we would see over 200,000 deaths in the period. (Anyone who wants to run a more rigorous stats analysis, please feel free to do so and send it to me.)

We often see a rise of 10,000 to 30,000 in the winter after a milder winter, so as winter 2018-19 saw 185,191 deaths, it’s quite likely that winter 2019-20 would see between 195,000 to 215,000 deaths. Looking at the raw figures back to 2006-07, 209,000 was the highest (in 2017-18), so we wouldn’t expect 215,000 unless it was a pretty bad winter, but getting over 200,000 was probable. As numbers in excess of 200,000 have been reached twice in the five-year period before this winter, that would not be anything out of the ordinary.

What actually happened in winter 2019-20?

So what did happen in winter 2019-2020 in England and Wales, the winter just gone? To work that out I added up the relevant weekly totals from the ONS spreadsheet, took away the days at the end of November and the start of April that were in there (assuming proportionate deaths for each day of the week), and got the answer 197,754.

(Note that ONS can’t calculate their EWD number for this winter until they’ve seen what the non-winter death figures will be like for some of this year, because they are used for calculating the winter excess. In fact, the official EWD figure for this winter is going to turn out to be very low, misleadingly low, because of the unusually high number of spring deaths – remember that EWD, as the ONS understands it, is calculated by a comparison to non-winter deaths.)

But some of that 197,754 are the Covid deaths that happened at the end of March, and we don’t want to include them because then we’d be double-counting when we come to look at the Covid numbers in the context of the winter deaths that had just happened. I calculate that there were 2131 Covid deaths in that number by 31 March, so our final number for winter 2019-20 deaths not including Covid is 195,623.

This is not, by recent standards, a particularly mild winter. There have been plenty of milder winters with deaths in 170,000s and 180,000s. So it is not true that we have had two mild (in the sense of relatively non-deadly) winters in a row. Nor, in fact, was 2018-19 unusually low when we look at in terms of the raw numbers, although it is in the lower half of usual range.

Subtract 20,000 deaths?

But on the other hand, 195,623 is at the lower bound of the range we decided was probable for winter 2019-20, which was 195,000-215,000. In other words, we could well have got another 20,000 deaths over the winter had we had a more dangerous flu going around. So we can conclude that it is somewhat probable that there were up to 20,000 elderly and/or medically vulnerable people who survived the December-March flu period, but who were at serious risk of death from any nasty viruses that happened to be around after that. The fact that they succumbed to Covid-19, which is a nasty disease, is, while sad, no great surprise, and also no cause for alarm, as these people could easily have died over winter when we would not have even noticed anything unusual happening, and we certainly would not have shut down the entire country for that.

So up to 20,000 of our 38,331 Covid deaths, over half, could have simply been deaths that could easily have happened anyway over winter, which would have been considered of no consequence for the overall health of the nation. Many of these people could have been expected to die by next winter anyway. I’m not saying that we shouldn’t care about such people and their slightly shortened lives, but considered from an objective, public health perspective, this is nothing to get worked up over.

If up to 20,000 of the Covid victims could have died anyway from an established disease like flu, then possibly only 18,000 Covid deaths can be considered ‘excess’ Covid deaths  (although it should have become clearer now that it is difficult to get any solid grasp on what we really mean by an ‘excess’ Covid death). And seeing as up to a half of Covid deaths may not really be Covid deaths, or could be deaths in which Covid only played a small part, perhaps hastening death by a few days or weeks, we possibly have very few ‘excess’ Covid deaths at all.

Sticking to 38,000

Even if you want to take a more hardline view, and claim that all 38,331 Covid deaths really are excess deaths, we still don’t have anything that major. A bad flu season can kill more people than that. (One of my next projects is to look into flu deaths more closely, unless someone else can do that.) Neither is a good thing, of course, but the point is that flu seasons have never worried us, and have not resulted in the insane measures taken by governments around the world, including the UK government. What should worry us are the excess, unprecedented, non-Covid ‘lockdown’ deaths, which are an absolute scandal, but which have been ignored by the media.

Isn’t it mostly ‘dry tinder’ anyway?

We also need to bear in mind that almost all Covid deaths are in very elderly people.

And elderly people with other serious conditions, serious enough to be mentioned on their death certificates as a probable contributor to their death.

These two graphs (from Christopher Bowyer) themselves tell us that some form of ‘dry tinder’ theory is right. It’s not like the 20,000 people I mentioned earlier were the only elderly people at risk of death from viruses that aren’t fatal for most other people. They’re the most vulnerable. But there’s hundreds of thousands of very old people in the country whose time is almost up. There are over three million people in the UK who are aged over 80, and they die off rapidly as they age – only about a third of those aged 80-84 make it to 90, and well over four-fifths of people aged 70-74 are dead before they reach 90. This may sound like I’m being callous, but I’m not. It applies, or will apply, to people I love as well. And me. And you. If you think you’re likely to make it to 100, think again. (You may do if you’re lucky, but you’re not likely to.) Besides, it’s just a fact. Like it or not, it’s true.

Some of these people are the ones who are also being killed by Covid. They’re the ones who die a bit before their time in the winters when an extra bad flu hits, or in the summers when a really bad heatwave takes place. I don’t want them to die when they might have had some more time left, but that’s a completely different thing than panicking that we have a major killer disease on the loose which is an existential threat to society. That would be a disease that kills large numbers of younger and fit and strong people. Covid-19 is not that disease, because it hardly kills any young people. Covid-19 mainly takes people who are in God’s waiting room. It is not a significant killer compared to other normal, established diseases like flu, it is in the same ballpark as them. It does not cause unusually large numbers of excess deaths of people who were not already near the end of their life. Governments should consider it as another flu-like illness, because that is what it is.

(Some might argue that without the lockdown vastly more people would have been killed, of course, but I think it’s very clear now that lockdowns are useless.)

Rick Hayward’s analysis

There are, of course, many other things to take into account with Covid, some of which have been published on this site in the last few months, but I would urge you to read, if you haven’t already, this Rick Hayward piece from July, because the focus of that is somewhat similar to this article, but it’s a far more useful and precise piece, and Rick also made the effort to work out the winter-spring mortality for the last 27 years. (Plus that piece should have got far more publicity than it did, it’s devastating, and I’m angry that it isn’t all over the media and the internet.)

Update: Please support my work by donating via Ko-fi, or subscribing via Patreon (see right-hand sidebar). Will sort out SubscribeStar soon for those who don’t like Patreon. Or buy my novel. This website receives no funding other than individual donations made by my brilliant readers, and I have been working full-time here and on Twitter for half a year now for virtually nothing to combat the tide of Covid misinformation and woke insanity.

Update 2: Another thing to note is that the EWD numbers show us that, far from being an unprecedented once-in-a-century mass killer, Covid-19 hasn’t even killed as many people as the cold and fogs of the 50s and 60s did. A population adjusted-graph would show this even more clearly. On that, see my historical post – another post which should be all over the media and the internet – here.

Appendix A:

The ONS’s EWD numbers (from here).

Winter	EWDs
1950-51	106400
1951 -52	44950
1952-53	82670
1953-54	47180
1954-55	64670
1955-56	67560
1956-57	29080
1957-58	57780
1958-59	77920
1959-60	41060
1960-61	68880
1961-62	68820
1962-63	89600
1963-64	49510
1964-65	41730
1965-66	57120
1966-67	25990
1967-68	70260
1968-69	50630
1969-70	67790
1970-71	34110
1971-72	46270
1972-73	46630
1973-74	34710
1974-75	31920
1975-76	58100
1976-77	34590
1977-78	41240
1978-79	48490
1979-80	30120
1980-81	32900
1981-82	42380
1982-83	42820
1983-84	30240
1984-85	47380
1985-86	49330
1986-87	26370
1987-88	32970
1988-89	21160
1989-90	47200
1990-91	37940
1991-92	34850
1992-93	25650
1993-94	25880
1994-95	27290
1995-96	40190
1996-97	47700
1997-98	22950
1998-99	46810
1999-2000	48420
2000-01	24790
2001-02	27230
2002-03	23930
2003-04	23420
2004-05	31570
2005-06	25190
2006-07	23540
2007-08	24620
2008-09	36330
2009-10	25660
2010-11	25960
2011-12	24040
2012-13	31150
2013-14	17280
2014-15	43720
2015-16	24500
2016-17	34410
2017-18	49410
2018-19	23200