Thursday 2 December 2021

COVID Follow-ups: The Four Countries of the UK

It’s been a few months since I last looked at COVID statistics. This time, I thought, I’ll get the figures for the four constituent countries of the UK, and present them against each other for comparison. I chose to look at cases, deaths and lockdowns, and to leave testing and vaccinations for later.

This exercise may also help to give some insights into the effectiveness, or otherwise, of various lockdown measures. This is because, for the last four to five months, England has been under a significantly lighter lockdown régime than Wales, Scotland or Northern Ireland.

Update, December 2nd: This is no longer the case, at least as far as face masks are concerned. See here: https://www.gov.uk/government/news/measures-against-omicron-variant-come-into-effect-30-november-2021. All this, so we are told, is due to a new variant, of which there have been a total of 22 cases in England and Scotland up to November 30th, compared with a UK new case count that day of almost 40,000. Yet again, the political class are missing no possible excuse to violate our rights and freedoms, even without any proven evidence. But, as I’ll explain towards the end, I think they’re going to find it hard to get away with this one.

The data

My usual data source, Our World in Data, doesn’t break the cases or deaths figures down to the level of individual constituent countries. I therefore had to go to the UK government’s website (https://coronavirus.data.gov.uk/details/cases and https://coronavirus.data.gov.uk/details/deaths) for these numbers. The data I used was taken on November 29th 2021, and runs up to November 27th (November 26th in the case of Wales).

I soon noticed some significant differences between this data and the UK-wide feed from Our World in Data. First, recent days’ deaths seem to be understated in comparison to the UK wide data. I suspect this may be due to deaths not being counted in individual countries’ data until they have been positively confirmed as being from COVID. Whereas in the UK-wide data, it looks as if an adjustment may be being applied to give a best estimate of the number of COVID deaths which will eventually be confirmed for that particular day. The same is true for the counts of daily cases, but the effects are smaller.

Second, the last few days’ data are not always complete; for example, there may be deaths reported for a recent day, for which there is no number of cases shown yet. I resolved this problem simply by not using the latest two days of data (three days in the case of Wales).

Thirdly, since I wanted to compare cases and deaths per million population, I had to find independent estimates of the populations of the four countries. I got a figure of 56,223,000 for England (estimated for July 1st 2021) from http://www.ukpopulation.org/england-population/. For Northern Ireland, I got 1,916,281, for Scotland 5,517,000 and for Wales  3,247,165, all from the same source. This gives a total of 66.9 million, as compared with the 68.2 million reported by Our World in Data. A bit odd; where are the 1.3 million missing people? Still, as I’ve said many times before, it’s the best data I’ve got, so I’ll use it.

Happily, the Blavatnik School of Government does break its lockdown measures figures down to the level of individual constituent countries; so, I can use those directly.

Cases

Here are the total cases per million:


The curves all have a similar shape, but there have been some divergences. In particular, Scotland’s case growth has fallen away relative to the others since September 2021. The UK as a whole is currently 15th among 190 world countries in cases per million, with just under 15% of the population having been reported as cases. As shown above, Northern Ireland has had a higher proportion of cases than this, and Scotland a lower one.

Here are the daily cases per million, weekly averaged:


The three major waves of cases in April 2020, October/November 2020 and January 2021 are clearly visible, though the middle one was far less prominent in Scotland. With regulatory divergence since June 2021, the comparison has become increasingly confused. Though the sudden almost vertical upsurge in cases, which was the “signature” of the delta variant, is evident in England in June to July, but hit Wales only in August, and Scotland seemingly in two phases, one as early as June, the other at the same time as Wales.

It’s also worth noting that the current daily cases per million rates in all four countries are well above the World Health Organization’s 200 cases per million per day threshold, at which they consider the virus “endemic” and advise countries not to unlock. At least, some of the politicians have shown more sense than unthinkingly obeying the WHO’s diktat. (But no longer: sigh).

Here are the weekly case growths since September 2020:

It’s clear that since June 2021 or thereabouts, the peaks and, to an extent, the troughs in weekly case growth (which is a very good proxy for the reproduction rate of the virus) have become more and more dampened. This is particularly noticeable in Wales, but is also visible in England (where the growth is less volatile, due to the larger population). In Northern Ireland and Scotland, the biggest recent peaks in weekly case growth came later, in July and August respectively.

Compared with earlier growth rates of over 100%, these are chicken feed. Suggesting that, absent new variants such as the “omicron” which is the supposed reason for the latest round of panic, the transmission of the virus now seems to be gradually, if irregularly, subsiding.

Lockdowns

Quantifying lockdowns is not as simple as it perhaps ought to be. The “official” published measure I use is the Blavatnik School of Government stringency index. This is a mixed metric of eight different lockdown types (schools, workplaces, public events, gatherings, public transport, stay at home, travel restrictions, international), plus a “public information campaign” metric. The stringency of each is assessed separately, depending on the level of lockdown and whether it is national or regional. Each is converted to a percentage, then they are averaged (all with equal weight) to give the stringency percentage.

Here are the stringency values over the course of the epidemic:


It looks as if, of the four, the Welsh have been most keen to micro-manage the situation as they see it. The Northern Irish started out by panicking less than the others; but in more recent months, as the other countries have unlocked, they have been reluctant to follow suit.

Since July, the English have been most eager to keep the stringency down. (Until now… grrr.) And judging by the total cases per million, in which England was overtaken by both Northern Ireland and Wales during that time, it doesn’t seem to have been a bad strategy. But politicians like Johnson don’t care about facts; nor about the people they are supposed to be serving. All they care about is virtue signalling and violating our rights and freedoms.

From my point of view, the Blavatnik stringency value has three drawbacks as it stands. First, the “public information campaign” metric is of little utility, since most countries in the world have kept it wound up to maximum practically throughout the epidemic. Propaganda is cheap, so they seem to think! Second, it doesn’t include face covering mandates. And third, it doesn’t really assess the experienced impact on our lives of the different lockdown measures.

Because of this, I created two metrics of my own. The first, which I simply call “lockdowns,” uses the same assessment of the stringencies of individual lockdown types as the Blavatnik index, but substitutes face covering mandates for public information campaign. And to avoid any confusion with the Blavatnik measure, instead of averaging the individual stringencies, it simply adds them together to give a score out of 900.

The second, which I call “harshness,” uses the same inputs as my lockdowns measure, but weights each type of lockdown according to how harshly it affects the lives of those impacted by it. Thus, lockdowns which affect everyone – like workplace closures (which affect both suppliers and customers), stay at home mandates and internal travel restrictions – are weighted higher than those which affect only some people, like school closures and international travel restrictions. And only mandates are included – mere recommendations are disregarded. Finally, to enable stringency and harshness to be shown together on the same plot, the individual harshness assessments are combined to form a percentage in the same way as the Blavatnik stringency measure.

Here is the current status of lockdowns, calculated according to my lockdowns measure:

At present, the Welsh, Scots and Northern Irish still have workplace restrictions, which the English do not. The Scots and the Welsh prefer to restrict public events rather than gatherings, whereas the Northern Irish restrict both, but gatherings more than public events. And the Northern Irish and Scots have (or had) more stringent face mask requirements than the English.

Two further graphs I’ll show are the average values of my lockdown metric over the whole course of the epidemic (weighted by number of days spent at each stringency level), and the proportion of the time spent in full lockdowns of each type (where “full” lockdown means 100% stringency according to the Blavatnik definition for that type of lockdown). Here they are:


These bar charts give an idea of which of the countries have chosen to lock down for the longest time, or to lock down hardest, for each of the lockdown types.

The effects of lockdowns on weekly case growth (and hence on reproduction rate) can be assessed using the following graphs, which plot case growth, stringency and harshness over time for each of the individual countries:




The times where lockdowns can be seen to have driven down weekly case growth to below zero are three. First, the initial lockdown in mid-March 2020, which produced falling new cases by about the second week of April (slightly later in Northern Ireland). Second, the brief lockdown of late October or early November. Though it’s worth noting that the case growth had come down on several previous occasions, even from big spikes, without a significant new lockdown. It’s also worth noting that this particular lockdown wasn’t needed at the time in Scotland. And third, the major lockdown of January to March 2021, which was released only slowly and bit by bit in the months from then until July. It’s debatable whether or not the Welsh lockdown in October 2021 actually had a significant effect on the case growth, or whether the spurt of cases would have died down anyway. And it’s debatable whether the Scottish lockdown of mid-October actually had any discernible effect on case growth at all.

It’s instructive, too, to compare the four countries’ lockdown harshness against each other. Harshness has generally a lower percentage value than the Blavatnik stringency; but the closer they are together, the more constraining the impact on individuals. It seems that the Northern Irish have tended to be less harsh, for a given level of stringency, than the rest.

Deaths

Here are the total deaths per million:


Now, that’s interesting. The English health “authorities” are well known to have reacted very poorly to the epidemic at the start, compared with other places. But since the second big wave of COVID deaths subsided around the end of March 2021, the other three countries have been, very gradually, catching up towards England. (Though I don’t think any of them is likely to get all the way there).

This is reflected in the graph of daily deaths per million (weekly averaged) – note, the strong drops at the far right of the graph are probably caused by deaths which haven’t yet been reported, since not yet confirmed as being from COVID:

England is maintaining a COVID daily deaths per million rate comparable with Wales and Scotland, and well below Northern Ireland, even though the English have been (had been?) locked down less stringently than the others for some months.

Here’s the graph of weekly deaths growth (again, the big drops at the right are probably due to deaths which haven’t been reported yet):

As with cases, the main trend over the last four months has been the decrease in the size of the peaks, and to a lesser extent the troughs, in weekly case growth.

Deaths per case

Here is the graph of deaths per case (with a 21-day offset) since September 2020. Again, the number of deaths on very recent days may be biased low:

Scotland and Northern Ireland have for at least six weeks been showing higher deaths per case than Wales or England. Perhaps there is a correlation between higher deaths per case and higher lockdowns? (I wouldn’t like to say whether there is causation, and if so which way!) In particular, the stricter face mask mandates in Scotland and Northern Ireland do not seem to have produced any corresponding reduction in recent deaths per case. (So, Johnson now wants to force people in England to wear face masks?).

I am loath to show the latest deaths per case histogram because of the drop-off issue, but the following graph based on the Our World in Data feed shows a current value of deaths per case of 0.35% for the UK as a whole, a value which has been close to constant for three months now:

Another way to look at deaths per case is to plot the cumulative deaths per case over the course of the whole epidemic:


Two things to note about the graph. First, the almost vertical drops in all the countries except Wales between June and July 2020. I presume these were due to adjustments, caused by re-definition of what constituted a COVID case and/or COVID death.

Second, that down-jag in the blue line in late February and early March 2020. There is evidence, from the world data for the very earliest phase of the epidemic, that the virus variant which left China in the middle of February 2020, after the (extended) Chinese New Year festivities, was something like two and a half times as lethal per case as the earlier one that had produced the first confirmed cases outside China. The interruption of this down-jag by an upward trend about 3 weeks after the new variant left China would seem to corroborate this. I myself am now glad that I caught the damned thing so early, that I got the original variant rather than the later, more lethal one – and passed it on to at least two friends!

So, where do we go from here?

It seems to me that Johnson and co are putting themselves in a cleft stick of their own making. The new face mask regulations in England are sure to raise a storm of protest from an already angry and disillusioned electorate. (I, for one, plan to boycott any business that refuses to serve me unless I wear a mask.)

But just as importantly, I don’t think they can win on this one. If the omicron variant fizzles, the entire excuse for their U-turn goes up in flames. If it explodes into dominance as the delta variant did, then making people wear face masks obviously won’t have done anything to help against it; so, they inconvenienced us under false pretences. If it persists, but makes little or no measurable difference to cases or deaths, then there won’t be a problem, and England will continue to do better than the other countries. So, why all the fuss?

To sum up

·       Early relative success (or, as in England, failure) in the early stages of the epidemic does not necessarily reflect what happens later on.

·       Contrary to received wisdom, lockdowns do not always have a discernible effect on cases or deaths. Indeed, the only three UK lockdowns which have shown such effects (in March 2020, October/November 2020 and January 2021 onwards) have all been large scale packages of measures, involving several different types of lockdowns at the same time.

·       Observed cases and deaths figures suggest that, in the absence of new and significantly more transmissible variants, the transmission of the virus is now tending gradually, if irregularly, to subside.

·       Whichever way things go with the “omicron” variant, I think Johnson and co have put themselves in a lose-lose-lose situation.

No comments: