by David Ogg
Amendment: This article was amended on 24 Feb to include the maximum 9% unexpected increase in deaths, in response to the discussion in comments below.
There has been much talk about the Covid death rate in Britain, with statistics and comparisons being thrown about like confetti. Luckily we don’t need to rely on opinion pieces and random stats because the Office for National Statistics (ONS) produces excellent data which we can study ourselves. This article is looking at the Age-Standardised Mortality figures for England and Wales, 2001 to 2020. (The Scottish figures won’t be available until June.)
- A warning: If you are uncomfortable with the fact that people die, look away now.
- A rider: I will be largely looking at the data for England as a whole. Welsh and Scottish data trends are usually similar but not the same, and likewise the regions of England differ but have similar trends.
Why is Age-Standardised Mortality such useful data? The number of people dying is a pretty solid statistic, and Age-Standardised Mortality is the best data for comparing across time (and countries) as it takes account of both population size and age profile. Conversely data on ‘Covid cases’ and ‘Covid deaths’ are both very difficult to tie down. For instance, as the ONS data shows, if the cause of death is to be taken at face value then in 2020 we had had a remarkable improvement in 6 of the 10 main causes of death. This doesn’t seem credible: it’s more likely that some deaths were misattributed to Covid.
So let’s look at the graph of Age-Standardised Mortality for England and Wales from 2001 to 2020:
Two things stand out immediately: There is clearly a sharp increase in 2020, but that is to a lower level than 2001 to 2008. So cause for concern, but not panic. The figures shown by the way are deaths per 100,000 people. In numbers, the level fell from 1230 in 2001 to 1084 in 2008, and jumped from 919 in 2019 to 1037 in 2020. (You can download the source data here.)
So let’s look into that 2020 increase. As is clear, it follows a particularly low year in 2019, the lowest ever in fact. There was a 3.9% drop from 2018 to 2019, so if 2020 had just gone back up to the 2018 figure that would have meant a 3.9% rise. There was actually a 12.9% increase, making an unexpected increase of 9% in 2020. To be clear, that’s 9% of the roughly 1% of the population who die a year, or 0.09% of the total population. However this would be a maximum: it may be stating the obvious but a dip in deaths one year tends to be followed by a corresponding bounce-back the next year, simply because that following year has its own usual cohort of frail and ill people, plus some of the previous year’s. If all of those frail and ill people who unexpectedly survived 2019 were to die in 2020, on top of the expected cohort, then we would anticipate a 7.8% increase from 2019 to 2020, ie 3.9% back up to the 2018 level plus the 3.9% who unexpectedly survived 2019. Given the actual 12.9% increase, that would give us a (minimum) unexpected increase of 5.1% in 2020. In summary, in 2020 there was an unexpected increase in deaths of between 5.1% and 9%, less than one in a thousand of the population (0.05-0.09%, about 35,000 to 60,000 deaths for the UK). The figures for Wales by the way are a 5.3% drop to 2019, 11.2% rise to 2020, so in fact a much smaller unexpected increase there.
If you are interested you can also see the detailed 2020 weekly data graph on the ONS site above. In practice in northerly latitudes it is almost invariably the severity of the seasonal ‘winter flu season’ (usually coronavirus or influenza) that creates these annual troughs and peaks, though in 2020 it happened a month or two later than usual with a nasty spike centred on April.
Let’s delve a bit deeper into the data trends. You can see that the graph above dips and stays down from 2016, with that year and 2017 and 2018 all rather low, followed by a very low year in 2019. So if we look at the average annual mortality for five-year blocks from 2001 we find something rather interesting: 2001-5: 1195, 2006-10: 1061, 2011-15: 973, 2016-20: 966. So the average mortality for the last five years, including 2020, is actually lower than the previous five, and a lot lower than the ten years before that. Looking at the data this way it would be reasonable to suggest that 2020 has simply taken up the slack from 2016-19. It’s a lesson in looking at the source data: It is quite reasonable for the media to compare 2020 to ‘the five year average’ but in this particular situation it conceals as much as it reveals.
The above is, I hope, solid data analysis. The following are some of my thoughts around this.
Age-standardised mortality data is reliable and allows realistic analysis across time, and the above conclusions in bold are two differing but reasonable interpretations of this data. Neither of them could be used to justify the enormous dislocations of lockdowns. The (lack of) effectiveness of the lockdown experiment has been dealt with extensively elsewhere, but the government’s published strategy for dealing with flu pandemics up to 2019 (which was designed to deal with far worse death rates than these by the way) would seem to have been much more appropriate. It’s worth a look – this coronavirus has in fact behaved much like previous coronaviruses and influenzas, and so much scientific knowledge and understanding was simply thrown away for this pandemic.
And since we know that the NHS has a crisis every winter, then increasing beds, staffing and ICU capacity would be a good way to go. If a system is running on the edge of crisis all the time it’s hardly surprising when it goes into one. And since the new Covid arrangements in hospitals (completely different from normal flu-management) have slashed bed numbers and staffing it could hardly be avoided. This looks like criminal long-term underfunding and mismanagement, and all the talk of saints, heroes and angels does no service to the staff or patients. The government has spent three times the annual NHS budget on its Covid strategy. Maybe building an NHS that could deal with fluctuations in demand would have been a better use of the money?
Future Mortality Trends
All things being equal we’d expect the age-standardised annual mortality figure to drop back to 970 or so in 2021 or 2022, i.e. the average of the last ten years or so, and carry on around there. However all things are not equal. It is accepted that the lockdowns and restrictions will cause large numbers of deaths which will emerge in the years to come. If annual age-standardised mortality really does drop to under 1000 we will have got away lightly.
And if I may, I will end with a word in favour of scepticism. Scepticism, that is questioning things and using our own judgement, is essential, most especially in politics and science. ‘Science’ without critical thinking is apologetics, not science: it is a fig-leaf for the powerful.
David Ogg is a retired public sector Project Manager